Fisheries science JSFS , tập 77, số 1, 2011 1

Vitellogenin In addition to being the precursor of ovarian lipovitellin, crustacean vitellogenin is considered to be an important transporter of lipids to the ovary from the hemolymph du

R E V I E W A R T I C L E

Mechanisms and control of vitellogenesis in crustaceans

T Subramoniam

Received: 13 May 2010 / Accepted: 7 October 2010 / Published online: 16 November 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract Crustaceans produce complex yolk proteins to

meet the substrate and energy requirements of embryonic

development Early electron microscopic investigations

point to a biphasic yolk synthesis, first within the ovary,

followed by heterosynthesis at extra-ovarian sites Recent

advances in molecular techniques have enhanced our

understanding of the genetic control of yolk synthesis in

crustaceans Amino acid sequencing of crustacean

vitello-genin (Vg) has enabled the elucidation of the cDNA

sequence associated with it, the identification of genes, and

the examination of their expression patterns in different

tissues Yolk processing in crustaeans involves cleavage of

the pro-Vg at consensus sites by subtilisin-like

endopro-teases within the hepatopancreas, hemolymph and oocytes

The structural elucidation of crustacean yolk proteins, as

well as the comparison of amino acid sequences of

vitel-logenins from various crustacean species, has revealed

molecular phylogenetic relationships not only among them

but also with other large lipid transfer lipoproteins of

dis-parate function The combinatorial effects of eyestalk

neuropeptides and a variety of trophic hormones achieve

the hormonal coordination of molting and reproduction

Biogenic amines secreted by the central nervous system

may also play an integrative role by stimulating

neuro-peptide secretion

Keywords Vitellogenesis Vitellogenin receptor 

Yolk processing Neuropeptides  Methyl farnesoate 

Ecdysteroids 17b-Estradiol

IntroductionMany malacostracan crustaceans produce large numbers ofyolk-laden eggs and brood them externally for extendedperiods Hence, vitellogenesis, the process of yolk forma-tion, is central to oogenesis In Crustacea, vitellogenesis is

a biphasic event consisting of autosynthesis and synthesis [1] This contention is supported by recentmolecular studies demonstrating yolk protein geneexpression both in ovary and hepatopancreas Receptor-mediated endocytosis of the yolk precursor molecule,vitellogenin (Vg), into growing oocytes has been estab-lished in crustaceans [2] The molecular transformation of

hetero-Vg into final yolk products for deposition in the matureoocyte is another crucial event in vitellogenesis

A defining feature in the endocrine regulation of logenesis in Crustacea is the occurrence of inhibitory hor-mones in the neurosecretory cells of the X-organ/sinus glandcomplex within the eystalk Conversely, many hormonalfactors as diverse in nature as methyl farnesoate (MF) andvertebrate steroidal hormones have been implicated in thestimulation of vitellogenesis However, we are far fromhaving achieved a clear understanding of the exact regula-tory mechanisms relating to the vitellogenic processes inCrustacea, mainly because of the species-specific nature ofthe effector molecules Yet, recent molecular studies on theprimary structure of the major vitellin molecules, as well asthe deciphering of their gene sequences and the elucidation

vitel-of their synthetic sites, are paving the way to an standing of the transcriptional control of the vitellogeningene in light of what is already known about insect andvertebrate vitellogenesis Homology searches and molecularphylogenetic analysis of various crustacean Vgs haverevealed unexpected results on their closer relationship withseveral members of the large lipid transfer lipoprotein

under-T Subramoniam ( &)

Marine Biotechnology Division,

National Institute of Ocean Technology, Velachery,

Tambaram Road, Pallikaranai, Chennai 600 100, India

e-mail: thanusub@yahoo.com

DOI 10.1007/s12562-010-0301-z

superfamily as compared to their own orthologous Vg

molecules This review undertakes a critical analysis of the

various mechanisms involved in the vitellogenic process and

their hormonal control

Vitellogenesis

Molecular composition of crustacean yolk proteins

Crustacean yolk proteins, referred to as lipovitellin, are

complex molecules comprising a high-density lipoprotein

(HDL) conjugated to carbohydrates and carotenoid

pigments [3] Crustacean lipovitellin differs from that of

vertebrates in that it lacks protein phosphates and has high

lipid content In the mole crab Emerita asiatica, purified

lipovitellin contains neutral lipids, glycolipids and

phos-pholipids, among which phospholipids are the dominant

lipid class, with phosphatidyl choline and phosphatidyl

serine being the major species [4,5] However, the

propor-tion of lipid to protein seems to be higher in the precursor

protein, Vg Crustacean lipovitellin characteristically

con-tains a variety of carotenoid pigments They include

beta-carotene, astaxanthin, canthaxanthin and cis-canthaxanthin,

among other minor intermediary metabolites [1] Crustacean

lipovitellin also possesses a higher carbohydrate content

than vertebrate vitellins In E asiatica, most protein-bound

carbohydrates found in lipovitellin are hexosamines and

hexoses [5] Emerita lipovitellin also contains

galactos-amine as well as O-linked oligosaccharides with N-acetyl

hexosamine as the terminal residue, whereas sialic acid is

specifically absent Khalaila et al [6] have identified the

glycosylation sites in the vitellogenin of the crayfish Cherax

quadricarinatus and characterized the glycan moieties

Besides providing an important source of carbohydrates for

the developing embryos, the glycosylation of Vg has an

important role in the folding and subunit assembly of these

molecules The glycan moieties may also play an equally

important role in the recognition of the Vg membrane

receptor during yolk accumulation After uptake into the

oocytes, they may also be involved in packaging and

com-pressing the yolk precursor proteins into the yolk bodies [5]

Biogenesis of yolk

In Crustacea, vitellogenesis occurs in two stages: a primary

vitellogenesis or previtellogenic phase characterized by the

differentiation of endoplasmic reticulum and the formation

of endogenous yolk stored in vesicles; and a secondary

vitellogenesis corresponding to an intensive phase of

uptake and storage of exogenous yolk precursor molecules,

which accumulate into large yolk globules [7] Early

electron microscopic investigations point to this biphasic

yolk synthesis, first within the ovary, followed by a erosynthetic yolk formation in somatic tissues such as thehepatopancreas or subepidermal fat body [1]

het-Support for autosynthetic yolk formation came from invitro incubation studies using ovaries of crayfish Pro-cambarus sp and of the crab Pachygrapsus crassipes [8]

In the shrimp, the yolk content of the egg is meager, andhence oocytes may be in a position to synthesize most ofthem, with only a very limited contribution deriving fromextra-ovarian sites In the kuruma prawn Marsupenaeusjaponicus, under in vitro conditions, only the ovaryincorporated radioactive amino acids into a proteinimmunologically identical to lipovitellin [9] Similar invitro studies on the vitellogenic ovaries of another shrimpPenaeus semisulcatus also revealed that the ovary is theprimary organ of vitellin synthesis [10]

Egg maturation in penaeid shrimp is characterized byvitellogenesis and cortical rod protein (CRP) formation In

M japonicus, Kim et al [11] showed that CRP mRNA ishighly expressed before the onset of vitellogenesis and that

Vg mRNA exhibited high expression during intense logenesis, suggesting that different genes are involved inthe ovarian synthesis of CRP and Vg proteins Okumura

vitel-et al [12] provided further evidence that eyestalk ablationinduced both Vg and CRP synthesis within the ovary.Khayat et al [13] demonstrated high levels of Vg mRNA inthe vitellogenic ovary of P semisulcatus, as evidenced byits ability to direct the cell-free synthesis of large amounts

of Vg However, unlike the other decapods, where synthesis of yolk has been shown to occur within theoocytes [8], in penaeid shrimp, the ovarian synthesis ofyolk probably takes place in the follicle cells Thus, in

auto-M japonicus, an immunofluorescence study with vitellin IgG was suggestive of yolk protein synthesis by thefollicular epithelium rather than by the oocytes Northernblot analysis and in situ hybridization have revealed thatmRNA encoding vitellogenin was expressed in the folliclecells of the vitellogenic females [14] Tsang et al [15] alsoshowed the expression of the vitellogenin gene, MeVg1, inthe ovary and hepatopancreas of Metapenaeus ensis, sug-gesting equal contributions from both tissues

anti-In recent years, several gene expression studies, usingquantitative real-time PCR techniques, have demonstratedthat the ovary remains the principal organ that synthesizesyolk proteins in several penaeid shrimp species Interest-ingly, in species such as M japonicus and P semisulcatus,one and the same Vg is expressed in the ovary and hepato-pancreas [16, 17] However, in other species, such asLitopenaeus merguiensis, M ensis and P monodon, morethan one Vg may be involved in the tissue-specific expres-sion of the gene in both the ovary and hepatopancreas[18–20] Especially in L merguiensis, the patterns of VgmRNA expression between the hepatopancreas and ovary

differ in that the expression level in the hepatopancreas is

much lower than that in the ovary at all stages of ovarian

development [18] Evidently, the relative contributions of

the ovary and hepatopancreas to overall yolk production

may differ among various shrimp species

Vitellogenin

In addition to being the precursor of ovarian lipovitellin,

crustacean vitellogenin is considered to be an important

transporter of lipids to the ovary from the hemolymph during

vitellogenesis In general, lipid transport through the

hemo-lymph is accomplished by two HDLs and a very high-density

lipoprotein (VHDL) [21,22] Female-specific vitellogenin is

one of the HDLs, with its production being correlated with

ovarian development in female crustaceans, whereas the other

HDL as well as VHDL are found in both males and females

In the penaeid shrimp, P semisulcatus, the non-sex-specific

hemolymph lipoprotein, LP I, consists of one 110-kDa

pep-tide unit, whereas the sex-specific LP II consists of 3 subunits

of 200, 120, and 80 kDa [23] Interestingly, the same subunits

were also present in the lipovitellin of this shrimp

Further-more, the lipid compositions of these two HDLs in P

semi-sulcatus also differ: LP II (Vg) has a lower lipid content than

does LP I, in addition to differences found in lipid classes

linked to the apolipoprotein Apparently, vitellogenin and

lipovitellin have similar protein structures, but show

differ-ences in their lipid contents, with the lipovitellin having more

percentage lipid acquired through adsorption within the

oocytes LP I is also different from LP II in its protein

com-position, as the former does not cross-react with anti-vitellin

antiserum In the crayfish Cherax quadricarinatus, Yehezkel

et al [24] observed that the hemolymph lipoprotein II,

equivalent to Vg, appears only at the onset of secondary

vitellogenesis In the mole crab E asiatica, Subramoniam and

Gunamalai [25] have described three hemolymph

lipopro-teins: LP1, LP2, and LP3 LP1 is non-sex-specific, but is

accumulated into the oocytes along with LP2, which is the

female-specific Vg LP3, which appears only during the

premolt of male and female crabs, plays a role in the transport

of lipids to the epidermis for the purposes of cuticle

forma-tion In addition to transporting a variety of lipophilic

compounds such as triglycerides and phospholipids,

crusta-cean Vgs are known to transport steroidal hormones like

ecdysteroids and vertebrate steroids, including estradiol 17b

and progesterone [26,27] These hormones are stored within

the oocytes as conjugates of yolk proteins and serve

regula-tory functions during embryogenesis

Site of vitellogenin synthesis

Initial electrophoretic and isotope tracer studies have

implicated several organs such as the hemocytes in crabs

[28, 29], the fat body in isopods and the amphipods[30,31], and the subepidermal adipose tissue in Palaemonserratus [32], and Scylla serrata [33] as the synthetic sites

of Vg However, the hepatopancreas has proven to be themost important organ synthesizing Vg outside of the ovary

in the majority of crustacean species analyzed The tacean hepatopancreas is the functional homolog to the fatbody in insects and the liver in vertebrates Subsequentinvestigations employing molecular techniques haverevealed that the hepatopancreas is the sole site of Vgsynthesis in the giant freshwater prawn, Macrobrachiumrosenbergii Chen et al [34] cloned a cDNA fragmentencoding Vg in this species and found its expression in thehepatopancreas of the vitellogenic female In addition,Yang et al [35] obtained cDNA fragments for four vitel-lins; using these cDNA fragments as probes, they found theexclusive expression of Vg mRNAs for the four vitellins inthe hepatopancreas of vitellogenic female M rosenbergii.Using quantitative real-time PCR techniques, Jayasankar

crus-et al [36] measured the expression levels of mRNA in thehepatopancreas of this species and also determined Vglevels using enzyme immunoassay Vg mRNA expression

in the hepatopancreas and hemolymph Vg levels showed agradual increase concomitant with increasing gonadoso-matic index Vg mRNA expression was, however, negli-gible in the ovary, confirming that the hepatopancreas isthe principal site of Vg synthesis in M rosenbergii Ingeneral, Vg expression may occur at multiple sites, butexpression patterns nevertheless vary according to species.That one and the same gene for vitellin and Vg can besimultaneously expressed both in the ovary and hepato-pancreas was shown in P semisulcatus [37] Multiplegenes may also show tissue-specific expression of Vg in theovary and hepatopancreas, as demonstrated in anotherpenaeid shrimp, Metapenaeus ensis, where two Vgs(MeVg1 and MeVg2) have been identified [15] TheMeVg1 gene is expressed equally in the ovary and hepa-topancreas, whereas MeVg2 is expressed only in thehepatopancreas Furthermore, the MeVg2 gene gives rise tosmaller transcripts, resulting in the production of manysmaller MeVg2 subunits destained for ovarian uptake [19].Evidently, the ovary is the primary site of yolk synthesis

in penaeid shrimp, as indicated by gene expression studiesenumerated above; on the contrary, large-bodied decapodssuch as crabs and lobsters seem to rely largely on extra-ovarian organs such as the hepatopancreas for the synthesis

of Vg Using molecular techniques, Li et al [38] havedemonstrated that in the Chinese crab Eriocheir sinensis,the hepatopancreas is the main site of Vg synthesis,although immunocytochemical studies have suggested aparallel role for ovary However, in the red crab Charybdisferiatus, northern blot analysis revealed that the crabexpresses the Vg precursor only in the hepatopancreas

In addition to the major 8.0-kb transcript, a large

propor-tion of smaller C feriatus Vg-specific transcripts are also

detected in the hepatopancreas These transcripts most

likely result from the alternative splicing and alternative

use of promoter and/or termination signals [39] The

occurrence of many Vg subunits in the crab hemolymph

may also result from autoproteolysis due to intrinsic

pro-tease activity in Vg itself [40] In a recent study using

quantitative real-time PCR techniques, Zmora et al [41]

found evidence that Vg is primarily expressed in the

hepatopancreas of the vitellogenic females, with only

minor expression in the ovary of the blue crab C sapidus

Furthermore, Vg expression in the hepatopancreas of this

brachyuran anecdysic crab is correlated with ovarian

mat-uration, with a remarkable 8000-fold increase in expression

from stage 3 to 4 of ovarian development Recent cloning

and expression studies on the Vg in the lobster Homarus

americanus also adduced further evidence that the

hepa-topancreas is the primary organ for yolk precursor

syn-thesis in lobsters [42] The lobster HaVg1, expressed

mainly in the hepatopancreas, comprises 14 introns and 15

exons This study also revealed that the sizes and locations

of the exons and introns of Vg are conserved among

crustaceans The HaVg1 precursor contained the

lipopro-tein domain at the N-terminus, followed by a domain of

unknown function in the middle The von Willebrand

factor type-D domain is located at the C-terminus of the

precursor A unique feature of crustacean Vg is that it

contains several cleavage sites, resulting in increased

subunit composition More numbers of Vg subunits may

also arise from smaller transcripts, as reported for the crab

C feriatus [43]

All these studies lead to the compelling conclusion that

the hepatopancreas is the principal site of Vg synthesis in

brachyuran crabs, lobsters, and probably other

representa-tive species under the suborder Pleocyamata Conversely, in

Dendrobranchiata, including mainly the penaeid shrimp,

both the hepatopancreas and ovary provide equal

contribu-tions towards Vg synthesis The Vg synthesized at

extra-ovarian sites such as the hepatopancreas undergoes several

modifications, such as glycosylation and lipid addition,

bringing about changes in molecular weight when compared

with the final yolk products accumulated within the ovary

[41] To sum up, besides being the precursor protein

mole-cule that supplies the amino acid pool for the developing

embryo, vitellogenin can also serve other subfunctions, such

as the transport of a variety of organic and inorganic

mole-cules required for embryonic development

Phylogenetic analysis of crustacean vitellogenin

Crustacean Vg is a multidomain apolipoprotein that is

cleaved into distinct yolk proteins Multiple alignments of

all known crustacean Vg sequences have revealed almostsimilar cleavage sites ClustalW alignment of M rosen-bergii Vg with that of 17 other crustacean species hasshown that the first common cleavage site RXRR occurs atamino acid residues 707–710, and the homology for thefirst segment is high when compared with the rest ofthe module The results from BLAST searches indicate thatthe N-terminal region of crustacean Vgs is conserved, as inthe apolipoproteins that are involved in lipid transport Thisproperty is in accord with the fact that Vg, insect apo-lipophorin II/I, apoB, and MTP are members of the samemultigene superfamily of large lipid transfer proteins(LLTP) [44] Next to the N-terminal segment, the middlesegment is comparable to a lipovitellin domain of unknownfunction called DUF1943 The C-terminal domain of

M rosenbergii Vg harbored a von Willebrand-factor type

D domain (YGP4) found in mammals Similarity in aminoacid sequence of the von Willebrand factor at the C-ter-minal region has also been reported for another LLTPprotein, the insect apolipophorin [45]

A phylogenetic tree constructed based on the alignment

of amino acid sequences of 18 crustacean Vgs using theClustalW programme shows six distinct lineage groups:Penaeidea (A), Brachyura (B), Astacidea (C), Caridea (D),Copepoda, and Brachiopoda (E), and Thalassinidea (F)(Fig.1; Table 1) The Vgs of the penaeidian species seem

to be highly homogeneous, with [92% identity in aminoacid sequence, except in the case of M ensis In M ensis,the two Vgs (MeVg1 and MeVg2), identified by Tsang

et al [15] and Kung et al [19], are expressed dently in the ovary and hepatopancreas, with only asequence identity of 56% between them These MeVgs alsoshowed less homology with Vgs of other penaeid shrimp,revealing a greater evolutionary distance from other pen-aeid species [46] As seen from Fig.1, Upogebia major,representing Thalassinidea, has taken a separate lineagenear the Brachyura In addition to all the above decapods,the two copepods and a branchiopod, Daphnia, formed aseparate clad in-between Caridia and Thalassinidea in theradial tree

indepen-The structural elucidation of Vg from different cean species has also been helpful in solving phylogeneticrelationships with other arthropod groups In a primitivebrachiopod, Daphnia magna, two Vgs, DmagVg1 andDmagVg2, have been isolated Interestingly, the lipidtransport module in the N-terminal region of DmagVg1 ismore closely related to those of insect Vgs than to those ofdecapod crustacean Vgs [47] Yet again, the intergenicregion of the two genes contains sequences resemblingjuvenile hormone-responsive and ecdysone-responsiveelements, typical of insect Vgs [48] The close homologyfound between Daphnia and insect LLT Vg modules may

crusta-be due to either divergence or convergence

In relation to other invertebrates, crustacean Vg has shown

homology in amino acid sequence with molluscan and coral

Vgs [49,50], although homology among the coral Vg, GfVg,

and the shrimp Vg (L vannamei) is much closer These

homology studies purportedly point to the emergence of Vg as

an egg protein precursor before the cnidarian–bilaterian

divergence The origin and evolutionary progression of Vgs

from a common ancestral molecule at the cnidarian–bilatrian

divergence denotes a landmark interception in crustacean

arthropods, giving rise to other lipid-carrying

apolipopro-teins that perform disparate physiological functions Even

among crustaceans, we find a number of lipoproteins, such

as crustacean clotting proteins and hemocyanin, that show

limited amino-acid sequence homology with vitellogenin [2]

The inclusion of crustacean Vg among other LLTP

proteins is also justified by the immunological relatedness

found between Vg of the crab S serrata and apoB, the

major protein component of LDL and VLDL [40] Warrier

and Subramoniam [40] demonstrated the recognition of Vg

by antibodies to apoB-containing mammalian lipoproteins

LDL and VLDL, and not to HDL (Fig.2) Furthermore, the

apoB antibodies reacted with greater efficacy to S serrata

Vg, thereby providing corroborative evidence for thestructural identity of apoB with Vg

Avarre et al [51] conducted a homology study betweencrustacean Vgs and other members of the LDL superfamily

of lipoproteins, and arrived at the conclusion that cean Vgs are closer to mammalian LDL and insectanapolipophorins However, the vertebrate apo-lipo B line ofproteins is thought to have diverged from the vertebrate Vgline, which, in turn, arose from the ancient egg yolk storageproteins of invertebrates [52] The closer relationshipbetween apoB and crustacean Vg discussed above notonly indicates the high conservancy in the lipid-bindingdomains of both these proteins, but may also point to theevolutionary derivation of vertebrate apo-lipo B proteins atthe crustacean Vg level

crusta-Vitellogenin receptors and yolk protein uptake

In crustaceans, only a few studies have been carried outwith reference to Vg receptors In the giant freshwater

Fig 1 Phylogenetics of eighteen crustacean Vgs with two dominant

domains, (A) penaeoidean and (B) brachyuran, at both ends of the

radial tree Other major domains such as Astacidea (C) and Caridea

(D) are found on either side of the tree Copepods and brachiopods

form a common grouping (E) along the brachyuran side idea (F) also has a separate lineage along the brachyuran crabs The protein sequence accession numbers for all the Vgs are given in Table 1

Thalassin-prawn M rosenbergii, Jugan and Soyez [53] demonstrated

Vg uptake by the oocytes by employing colloidal

gold-conjugated vitellin The labeling was visualized in the

microvilli, coated pits, and intraooplasmic vesicles These

authors further observed that a sinus gland neuropeptide

inhibited vitellogenin endocytosis, possibly by blocking themembrane receptors Laverdure and Soyez [54] solubilizedthe vitellogenin receptor from the oocyte membrane of

H americanus, and characterized it using an linked immunosorbent assay Binding of Vg with the sol-ubilized receptors increased at the onset of vitellogenesis,but decreased in older oocytes of the freshwater crayfishOrconectus limosus [55] The solubilized oocyte mem-brane receptor with a molecular weight of 28–30 kDabinds specifically to Vg of O limosus Warrier and Subr-amoniam [56] purified the vitellogenin receptor in the mudcrab Scylla serrata using HPLC and found a still highermolecular weight of 230 kDa In direct binding studiesusing 125I-labeled Vg, crab VgR was observed to haveincreased affinity to its ligand in the presence of Ca2?andwas inhibited by suramin, a polysulfated polycyclic

enzyme-Table 1 Vg accession numbers of 18 crustaceans, along with their taxonomic classifications

Penaeus monodon ABB89953.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Penaeus Fenneropenaeus chinensis ABC86571.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Fenneropenaeus Fenneropenaeus merguiensis AAR88442.2 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Fenneropenaeus Litopenaeus vannamei AAP76571.2 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Litopenaeus Marsupenaeus japonicus BAD98732.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Marsupenaeus Metapenaeus ensis AAT01139.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Dendrobranchiata;

Penaeoidea; Penaeidae; Metapenaeus Homarus americanus ABO09863.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Astacidea; Nephropoidea; Nephropidae; Homarus Cherax quadricarinatus AAG17936.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Astacidea; Parastacoidea; Parastacidae; Cherax Portunus trituberculatus AAX94762.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Brachyura; Eubrachyura; Portunoidea; Portunidae; Portunus Callinectes sapidus ABC41925.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Brachyura; Eubrachyura; Portunoidea; Portunidae; Callinectes Charybdis feriatus AAU93694.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Brachyura; Eubrachyura; Portunoidea; Portunidae; Charybdis Scylla serrata ACO36035.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Brachyura; Eubrachyura; Portunoidea; Portunidae; Scylla serrata Macrobrachium rosenbergii BAB69831.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Caridea; Palaemonoidea; Palaemonidae; Macrobrachium Pandalus hypsinotus BAD11098.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Caridea; Pandaloidea; Pandalidae; Pandalus Upogebia major BAF91417.1 Malacostraca; Eumalacostraca; Eucarida; Decapoda; Pleocyemata;

Thalassinidea; Callianassoidea; Upogebiidae; Upogebia Daphnia magna BAE94324.1 Branchiopoda; Diplostraca; Cladocera; Anomopoda; Daphniidae;

Daphnia Lepeophtheirus salmonis ABU41135.1 Maxillopoda; Copepoda;Siphonostomatoida; Caligidae; Lepeophtheirus Tigriopus japonicus ABZ91537.1 Maxillopoda; Copepoda; Neocopepoda; Podoplea; Harpacticoida;

Harpacticidae; Tigriopus; Tigriopus japonicus

Fig 2 Dot blot analysis of crab Vg (1), rat LDL (2), VLDL (3), and

HDL (4) using anti-crab Vg antibodies (dilution 1:2000) Anti-Lv

antibodies are seen to react well with Vg, LDL, and VLDL, but there

is no reaction with HDL (from Warrier and Subramoniam [ 40 ])

hydrocarbon These authors also showed an immunological

relatedness between VgR of S serrata and LDLR by virtue

of the ability of VgR to bind rat LDL and VLDL

In a recent study, the cloning and characterization of a

cDNA encoding a putative Vg receptor from the tiger

prawn P monodon (PmVgR) has been reported [57]

PmVgR has a molecular weight of 211 kDa, and is ovary

specific It consists of conserved cysteine-rich domains,

EGF-like domains and YWTD motifs, similar to the

mammalian LDL receptor as well as to the Vg receptors of

insects and vertebrates PmVgR expression in the ovary

coincides with the rapid pace of Vg production by the

hepatopancreas Immunological detection of PmVgR in the

oocyte membrane during intense vitellogenesis has also

been done in this prawn Further, PmVgR expression was

knocked down in animals after they were injected with

PmVgR dsRNA, leading to a decrease in vitellin content in

the ovary, and at the same time elevating the levels of

hemolymph Vg A similar molecular characterization of

VgR has also been reported for the kuruma prawn

M japonicus [58] The expression dynamics of MjVgR

during vitellogenesis have been found to be similar to those

of P monodon Furthermore, structural analysis of the VgR

of this shrimp also reconfirmed its inclusion in the LDLR

superfamily The results of these studies are comparable

with those of S serrata with respect to molecular weight

and functional characteristics [56] It would be of interest

to know whether crustacean VgR also facilitates the

endocytosis of other hemolymph lipoproteins into the

ovary, similar to avian VgRs [59] and insectan lipophorin

receptor [60]

Receptor-mediated internalization of Vg into the

oocytes has been demonstrated by an immunogold

elec-tron microscopic study using anti-Vg as the primary

antibody in S serrata [56] Immunogold labeling against

Vg antibody was first visualized in the coated pits found

on the plasma membrane of the vitellogenic oocytes This

is followed by their appearance in the pinched-off coated

vesicles as well as in early endosomes, which fuse

toge-ther to form the mature electron-dense late endosomes

(Figs.3, 4) Such an endocytotic entry of Vg into the

oocytes to form the yolk body is similar to that described

for insects [61] In P monodon, after the binding of Vg

with VgR, the complex moves into the oocyte cytoplasm,

aided by internalization signals present in VgR [57]

Interestingly, the VgR of P monodon has two putative

internalization signals (i.e., FANPGFG and FENPFF)

found in vertebrate VgRs as well as several IL and LI

sites characterizing the insect VgR and Drosophila yolk

peptide receptors [57] This redundancy with the

inter-nalization signals present in the shrimp oocytes could

increase the efficiency of receptor-ligand binding during

crustacean vitellogenesis

Yolk processing

In general, Vg undergoes post-translational proteolyticcleavage at the site of synthesis (e.g., insects [61]) or aftersequestration into the ovary (e.g., amphibians [62]) In crus-taceans, SDS-PAGE analysis of hemolymph and ovary yolkproteins has indicated the occurrence of varying numbers of

Vg and vitellin (Vn) fractions, suggesting that Vgs are alreadyfragmented at the time of endocytotic uptake into the ovary

In the isopod Armadillidium vulgare, four female-specificglycoprotein bands in hemolymph, detected on SDS-PAGE,

Fig 3 Immunogold labeling of Vg in ultrathin sections of the ovary

of Scylla serrata, examined with a Philips CM10 transmission electron microscope to demonstrate the endocytosis of Vg Vg labeling is seen along the luminal surface of the coated vesicle (cv), which fuses into a mature endosome Electron-dense particles representing Vg molecules are densely packed within the endosomes Scale bar 0.5 lm (from Warrier and Subramoniam [ 56 ])

Fig 4 Immunogold labeling of Vg in ultrathin sections of the ovary

of Scylla serrata In this micrograph, fusion of an early endosome (ee) with a mature endosome is observed (indicated by an arrow) Electron-dense particles of Vg are extensively labeled in the mature endosome compared to the early endosome The mature endosomes finally form the yolk bodies Scale bar 0.5 lm (from Warrier and Subramoniam [ 56 ])

were found to be the same in the ovarian extract [63] In this

isopod, an anion-exchange HPLC separation has yielded 6

vitellins from the ovary, ranging in molecular weight from

112 to 205 kDa [64] The N-terminal sequencing of these

proteins showed identical amino acids except for the 112 and

59 kDa proteins PCR-assisted cloning of the 50 region of a

cDNA encoding Vg revealed the presence of an

amino-ter-minal sequence identical to those of the 112 and 122 kDa yolk

proteins, suggesting that the Vg gives rise to the Vn fractions

by cleavage either in the hemolymph or ovary

In M rosenbergii, Vg, after being synthesized as a single

precursor protein, undergoes initial cleavage at amino acids

707–710 by a subtilisin-like endoprotease to give rise to two

subunits, A and pro-B, within the hepatopancreas [65] After

secretion into the hemolymph, subunit A is sequestered as is

into the ovary, whereas pro-B is cleaved by another processing

enzyme to give rise to subunits B and C/D (Fig.5) The ovary

subsequently takes them up to give rise to the yolk proteins,

VnA, VnB, and VnC/D Examination of subunit composition

of Vg in hemolymph and Vn into the ovary by SDS-PAGE and

western blotting has also supported the above sequence of Vgconversion to Vn fractions Furthermore, identity in theN-terminal amino acid sequences of these Vg and vitellinfractions that appear in hemolymph and ovary has also pro-vided final support to the scheme of Vg processing in thisfreshwater prawn [65] Further studies on the processing ofother decapod crustacean vitellogenins have revealed conser-vancy in the first cleavage site at amino acids 707–710,although the subsequent cleavage sites may differ amongmany species In Litopenaeus vannamei, Raviv et al [66]predicted an N-terminal sequence of 78 kDa, with the firstcleavage site occurring at an RTRR consensus cleavage forsubtilisin-like endoprotease These authors isolated five HDLpolypeptides of masses 179,113, 78, 61, and 42 kDa from theovary and found that all of these polypeptides are derived fromthe 179 kDa second fraction of the premature Vg of L van-namei These results are in accord with those described for

M rosenbergii yolk protein processing In a recent study on themud shrimp, Upogebia major, belonging to the infraorderThalassinidea of Decapoda, Kang et al [67] found threepolypeptides in the oocytes These subunits were found to bederived from a single long polypeptide translated from the Vgtranscript in the hepatopancreas This precursor polypeptide of

289 kDa is cleaved to produce two Vg subunits at the sensus cleavage site, RLRR, which is recognized by subtilisin-like endoproteases These two subunits are also suggested toundergo further processing upon or immediately after incor-poration into oocytes

con-Evidence for the secondary cleavage of vitellogenin afterits uptake into the ovary is given in other decapods such asthe freshwater crayfish, Ibacus ciliates [68] A low-densitylipoprotein isolated from the ovary of this crayfish degraded

Vg into apolipoprotein fragments, which are similar to thelipovitellin subunits of the egg Furthermore, the Vg diges-ted by LDL exhibited proteinase activity whereas the native

Vg did not have it The instability of Vg and its susceptibility

to undergo proteolytic cleavage may be a general feature,but in a brachyuran crab Scylla serrata, Vg itself possessesproteinase activity [40] Warrier and Subramoniam [40]demonstrated that conformational changes in the native Vgcould bring about such proteolytic cleavage, as indicated in astudy using urea as a destabilizer Whereas Vg showed aspectral change with 8 M-urea treatment due to exposure ofthe hydrophobic core containing aromatic residues(absorption at 274 nm), lipovitellin did not show such aspectral shift Clearly, Vg is a relatively unstable lipopro-tein, but the ovarian lipovitellin is more stable

Yolk utilizationYolk proteins primarily evolved to supply both energy aswell as organic building blocks to support embryonic

Fig 5 Schematic representation of synthesis and processing of

vitellogenin in Macrobrachium rosenbergii Vg is synthesized as a

single precursor molecule, A–B–C/D, in hepatopancreas, which is

then cleaved into two subunits, A and proB Subunits A and proB are

released into the hemolymph, and proB is cleaved to form two

subunits B and C/D The three processed subunits A, B, and C/D are

incorporated into the ovary (From Okuno et al [ 65 ])

growth in oviparous animals Understandably, yolk

utili-zation is the central event of embryogenesis, and is

accomplished by a host of hydrolytic enzymes acting on

the complex yolk molecules Subramoniam [69] has

reviewed the existing information on crustacean embryonic

nutrition from the perspective of yolk utilization During

yolk utilization, the complex lipovitellins are dismantled

by esterases, proteases and glycosidases, resulting in the

release of conjugated steroidal hormones [70] The

regu-lated release of active ecdysteroids from their conjugates

by nonspecific esterases at specific times in embryogenesis

may not only trigger embryonic cuticle formation but may

also accomplish larval molting and egg hatching [26,71]

Direct utilization of lipovitellins in the egg by way of

proteolytic cleavage in the developing embryos has also

been documented in the blue crab C sapidus [72]

As much as the yolk proteins meet the metabolic

demands of embryonic development, they are also used in

early larval development In an extreme case of cirripede

development, a new protein is expressed during the

non-feeding cypris stage of the barnacles This protein, called

cypris major protein, is interestingly related to the heavy

chain of barnacle yolk protein both structurally and

immunologically [73] Another glycoprotein, called

set-tlement-inducing protein complex (SIPC), which is found

in juvenile and cyprid larvae of the barnacle Balanus

amphitrite, also showed immunological and peptide

sequence similarity with cirripede yolk proteins [74]

Evidently, cirripede larval storage protein and the SIPC

may share a common ancestor with yolk protein

Alterna-tively, crustacean yolk protein genes would have

under-gone duplication to give rise to different proteins necessary

for larval metamorphosis and gregarious larval settlement

in these sessile barnacles

Endocrine regulation of vitellogenesis

In most malacostracan crustaceans, except the diecdysic

crabs, vitellogenic activities are sandwiched between two

molt cycle stages Such an inextricable linkage between

molting and vitellogenesis is accomplished by a delicate

multihormonal interaction unique to crustaceans Egg

brooding within the pleopods of several malacostracans

provides another intervention in the coordinated control of

molting and reproductive cycles Essentially, the hormonal

controlling mechanisms enabling the temporal separation of

these two processes involve principally the inhibitory

neu-ropeptides—vitellogenesis-inhibiting hormone (VIH) and

molt-inhibiting hormone (MIH)—originating from the

X-organ/sinus gland complex in the optic ganglia Thus, the

hormonal coordination of both molting and vitellogenesis

becomes vital to accomplishing continued body growth and

increased fecundity [75] The endocrine factors that controlvitellogenesis can be considered under two categories:gonad-inhibiting and gonad-stimulating hormones

Gonad-inhibiting hormonesGonad-inhibiting hormones of Crustacea mainly reside inthe eyestalk X-organ/sinus gland complex The crustaceanhyperglycemic hormone (CHH) superfamily of neuropep-tides that mainly originate from this neuronal complexinclude important regulatory molecules to control somaticgrowth and reproduction CHHs themselves play a pivotalrole in the regulation of glucose metabolism However, theyalso exhibit considerable cross-functional activities withother peptides such as MIH, VIH, and mandibular organinhibitory hormone (MOIH) The application of peptidesequencing as well as PCR-based cloning techniques hasresulted in the isolation of many cDNA sequences of CHHfamily members involved in diverse regulatory functions Inaddition, these investigations have facilitated sequencehomology studies to establish structural relationships amongthem Their neuronal distribution outside eyestalk gangliaimplicates other parts of CNS such as supraesophagealganglia, thoracic ganglia and ventral nerve cord in the reg-ulatory roles of molting and reproduction

Vitellogenesis-inhibiting hormoneVitellogenesis-inhibiting hormone belongs to the CHHfamily of neuropeptides, and shows inhibitory effects onovarian growth and vitellogenesis Our present under-standing of endocrine regulation of crustacean vitellogene-sis per se is mainly based on experimental studies involvingthe removal of VIH by way of eyestalk extirpation VIH wasfirst characterized in the American lobster H americanus as

a 78-residue peptide that exists as two enantiometric forms, both of which have a molecular mass of 9135 Da, anamidated C-terminus and a free N-terminus [76,77] How-ever, the vitellogenesis-inhibiting effect was found in onlyone isoform when tested with an in vivo heterologous assaydeveloped in the grass shrimp Palaemonetes varians VIHhas been subsequently isolated and characterized from manymalacostracans, and has been shown to play a prominent role

iso-in the regulation of reproduction, especially vitellogenesis[78] Amino acid sequence homology studies on the VIH ofseveral crustacean species have uncovered considerablesimilarities with other CHH family peptides such as MIHand MOIH, claiming a separate subgroup (Type II) from theCHH molecules [79]

Bioassay studies to test VIH activity have been carriedout either using an ovarian growth index [80, 81] or by

in vitro culturing of ovarian tissue and monitoringthe inhibition of protein synthesis [82, 83] Inhibition of

gold-labeled vitellin binding to oocyte microvilli in

incu-bation medium containing sinus gland extract is another

bioassay method that was followed by Jugan and Soyez

[53] Another in vivo bioassay system involving the

mea-surement of vitellogenin levels in the hemolymph by a

highly sensitive sandwich enzyme-linked immunosorbent

assay was employed to quantify Vg in eyestalk-ablated P

monodon [84] In this species, only two of the

HPLC-purified eyestalk peptide fractions were found to reduce

hemolymph Vg concentrations in a time-dependent manner,

suggesting their direct inhibitory effect on Vg synthetic

sites In H americanus females, the highest hemolymph

levels of VIH were observed during the immature and

previtellogenic stages [85] Edomi et al [79] isolated two

VIH sequences from the eyestalk of the Norway lobster

Nephrops norvegicus Interestingly, mRNA expression of

VIH in this lobster was detected not only in the eyestalks

but also in the supraoesophageal ganglia In a recent study

using double-stranded RNA (GIH-dsRNA), Treerattrakool

et al [86] knocked down GIH expression both in the

eye-stalk ganglia and abdominal nerve cord in P monodon This

resulted in a conspicuous increase in Vg transcript level in

the ovary of GIH-knockdown shrimp, although Vg

expression in the hepatopancreas was less significant

The inhibitory effects of eyestalk hormones with

par-ticular reference to VIH were further investigated by

Tsutsui et al [16] in another penaeid shrimp species,

M japonicus Using a quantitative real-time PCR system,

Vg mRNA expression levels were measured both in the

hepatopancreas and the ovary in normal and

eyestalk-ablated adult shrimp Their study indicated a significant

increment in mRNA levels in the ovary but not

hepato-pancreas, suggesting that VIH exerts its effects primarily

through vitellogenin gene expression in the ovary

Hepatopancreatic gene expression may not be significantly

affected by VIH, although in this shrimp, vitellogenin

cDNA from the hepatopancreas is identical to that isolated

from the ovary [14] In a more recent study, Tsutsui et al

[87] isolated as many as six sinus gland peptides with

vitellogenesis-inhibiting activities in the whiteleg shrimp

L vannamei These VIHs caused varying degrees of

inhi-bition in Vg mRNA expression in ovarian fragments of

M japonicus incubated in vitro Marco et al [88] have

predicted that the presence of a C-terminal amide in two

CHHs of J lalandii could be responsible for VIH activity,

based on tests done using a P semisulcatus ovarian

incu-bation system Recently, Ohira et al [89] produced a

recombinant VIH from the American lobster H americanus

and tested its inhibitory activity on ovarian fragments of

M japonicus in a culture system The amidated C-terminus

of this recombinant neuropeptide has also been shown to be

responsible for its vitellogenesis-inhibiting activity In

M japonicus, cyclic nucleotides such as cAMP and cGMP,

Ca2?, and protein kinase C appear to serve as second sengers in mediating Vg mRNA synthesis in the ovary [90].Cyclic AMP and cGMP probably mediate the action of VIH

mes-on Vg synthesis in the follicle cells of the ovary In thisshrimp, the responsiveness of the ovary to VIH ishigh during previtellogenesis, compared to the vitellogenicovary [91]

Eyestalk ablation affecting Vg synthesis has also beendemonstrated in the hepatopancreas of the giant freshwaterprawn M rosenbergii In the adult female, Vg mRNAexpression increases significantly in the hepatopancreas,with concurrent elevations in hemolymph Vg levels as well

as gonadosomatic index [36] More significantly, suchincreases in mRNA levels in the hepatopancreas, increasedlevels of Vg in the hemolymph, and elevated gonadoso-matic index have been shown in eyestalk-ablated juvenilefemale prawns [92] In the eyestalk-less isopod Armadil-lidium vulgare, VIH suppressed Vg synthesis in incubatedfat body tissues [93] Taken together, the above results areindicative of the fact that VIH acts on the target tissuessuch as ovary, hepatopancreas and fat body that areinvolved in Vg synthesis in all species investigated thus far.Androgenic gland hormone

Vitellogenin is a female-specific hemolymph protein, and

in a sense, can be said to be a secondary sexual teristic in the reproducing female In many submammalianvertebrates such as amphibians and fishes, Vg expressioncan be induced in the male liver by exogenous estrogen[94] In crustaceans too, Vg induction has been shown inthe fat body of androectomized male isopods [30, 95].These studies have demonstrated yet another Vg-inhibitingfactor that resides in the androgenic gland of male crusta-ceans The VIH-like effects of androgenic gland hormoneare well known in the protandric hermaphrodites In thehermaphroditic caridean prawn Pandalus hypsinotus, Vg isnot expressed in males and immature females, but becomesdetectable from the late male phase associated with thedegeneration of the androgenic glands and the appearance

charac-of vitellogenic oocytes in testicular tissues [96] In anothersexually plastic crustacean, the freshwater crayfish

C quadricarinatus, Vg is not expressed in intersex viduals, while transcription of the gene is induced in thehepatopancreas when the androgenic glands are removed[97,98] Evidently, the androgenic gland plays an essentialrole in negatively regulating the expression of the female-specific Vg gene in intersex individuals

indi-Mandibular organ inhibitory hormoneJust as the glandular Y-organ is controlled by MIH from theX-organ/sinus gland, and with the analogy of allatostatins

controlling juvenile hormone synthesis by the corpora allata

in insects, the synthesis of crustacean juvenoid, MF, in the

mandibular organs (MOs) is inhibited by an eyestalk

neu-ropeptide named MOIH [99] First described in the spider

crab Libinia emarginata, MOIH exists in three forms, all of

which repress MF synthesis to a degree of 70–80% [100]

However, the crude extract of the sinus gland showed more

than 90% inhibition, suggesting that there is a combined

effect of a group of neuropeptides from the eyestalk that

controls MF synthesis MOIH isoforms of L emarginata

have a molecular weight of 8,400 Da, while sharing other

features of CHH family peptides, including N-terminal

blockade by pyroglutamic acid [100] MOIH was also

characterized biochemically in the shore crab Cancer

pagurus by Wainwrite et al [101], who identified two

iso-forms (MOIH I and MOIH II) with almost identical amino

acid sequences, except for the replacement of Lys in MOIH I

with Glu at position 33 of MOIH II In this crab, the

sensi-tivity of MO to MOIH I is high at the beginning of

vitello-genesis and declines drastically during peak vitellovitello-genesis,

indicating a stage-specific role for MF on Vg synthesis

[102] The mechanism of MOIH action on MF synthesis

involves the inhibition of farnesoic acid

O-methyltransfer-ase (FAOHeT), the enzyme that catalyzes the final step of

MF biosynthesis in the MOs, by affecting the methylation of

FA to produce MF In C pagurus, high MF titers occur

before or during early vitellogenesis, and coincide with or

are preceded by elevated levels of putative FAOMeT mRNA

in the MOs [103] Sequence studies on MOIH I and II

peptides of C pagurus have revealed their close identity

with MIH and VIH, although none of the CHH peptides

exhibited MOIH activity in this crab [103] On the other

hand, in the spider crab L emarginata, all the three isoforms

of MOIH exhibited CHH activity when injected into the

eyestalk-ablated fiddlercrab Uca pugilator [100] Inhibition

of MF synthesis by MOIH and other CHHs assumes greater

physiological significance in view of the dual role that the

MOs play in the regulation of both reproduction and molting

in decapod crustaceans

Vitellogenesis-stimulating hormones

It is possible that crustaceans employ multiple hormonal

factors to positively control vitellogenesis They may be

species specific or combinatorial in action, and they are

varied in chemical nature They include (1) the

neurose-cretory hormones from the brain/thoracic ganglia, (2)

methyl farnesoate, a structural homolog of insect juvenile

hormone III, and farnesoic acid (FA), secreted by the

mandibular organs, (3) ecdysteroids, and (4) a variety of

steroidal hormones, including estrogen and progesterone of

uncertain origin Biogenic amines secreted from the central

nervous system also seem to play a pivotal role in thecontrol of female reproduction by influencing the secretion

of both gonad-stimulatory and -inhibitory neuropeptides.Although much experimental evidence exists to implicatethese hormonal factors in vitellogenesis, the action of thesehormones at the level of gene transcription in Crustacea isonly beginning to be understood

Gonad-stimulating hormonesThe first evidence for a gonad/vitellogenesis-stimulatingprinciple in the central nervous system of Crustacea wasobtained by Otsu [104], who noticed precocious ovariandevelopment in the crab Potamon dehaani after theimplantation of thoracic ganglia Following this discovery,several attempts have been made to implant brain and tho-racic ganglia or to inject their extracts to stimulate vitello-genesis in different crustacean species [105–107] In thiscontext, the role of differing biogenic amines in influencingthe release of neurosecretary peptides from differentneurosecretory neurons is relevant to understanding theirintegrative role in crustacean vitellogenesis That theadministration of serotonin (5-hydroxytryptamine; 5-HT) iseffective at stimulating ovarian maturation was indicatedfirst in the fiddler crab, Uca pugilator [108] Subsequently,Sarojini et al [109, 110] demonstrated in the freshwatercrayfish Procambarus clarkii that dopamine (DA) inhibits5-HT-stimulated ovarian maturation by inhibiting therelease of gonad-stimulating hormone (GSH) from the brain

or thoracic ganglia, or enhancing the release of VIH from theeyestalk neurosecretory centers The opposing effects of5-HT and DA on vitellogenesis were demonstrated in sev-eral other crustacean species, including the Indian spinylobster Panulirus homarus [111]; but in the giant freshwaterprawn M rosenbergii, Chen et al [112] provided experi-mental evidence that the site of action of DA is at the tho-racic ganglia through the inhibition of the release of GSH,and not by the enhancement of VIH secretion

Crustacean hyperglycemic hormones secreted from the

X organ/sinus gland complex of the American lobster

H americanus are shown to have multiple functions,including molt inhibition and gonad stimulation [113] Inthis lobster, CHH exists as two isoforms, CHH-A andCHH-B Interestingly, both GIH and CHHs are produced inthe same neuroendocrine cells mRNA levels as well asCHH titers in the hemolymph indicate that CHH-Bexpression in particular peaks during intense vitellogenesis[113] Furthermore, the hemolymph levels of GIH are highwhen CHH is low, and vice versa CHH-A and CHH-B arealso present in parts of the nervous system other than theoptic ganglia, raising the question of whether they are thesame substances as the so-called gonad-stimulating hor-mones of the brain and thoracic ganglia

Molt-inhibiting hormone

The cross-functional role of the CHH family of peptides is

also found with MIH of several decapod crustaceans [114]

MIH exists in two isoforms, MIH-A and MIH-B, in the

penaeid shrimp M ensis Interestingly, MIH-B is expressed

not only in the X-organ/sinus gland, but also in the ventral

nerve cord, thoracic ganglia, and brain during

vitellogen-esis The levels of MIH-B mRNA transcript in the eyestalk

decrease in the initial phase of gonad maturation and

increase towards the end of maturation, suggesting a

stimulatory role for this neuropeptide in the initiation of

vitellogenesis Further, the injection of rMIH-B delayed the

molting cycle of the maturing female [114] and increased

levels of Vg mRNA expression and Vg synthesis in the

ovary and hepatopancreas of this shrimp [115] Injection of

MeMIH-B dsRNA into female shrimp also caused a

decrease in MeMIH-B transcript levels in the thoracic

ganglia and eyestalks Similarly, in C sapidus, Zmora et al

[116] recently found that MIH titers are significantly higher

in the mid-vitellogenic stages rather than in the early

vitellogenic stages While high MIH levels during

inter-molt suppress inter-molt hormone synthesis by the Y-organs in

the anecdysic blue crab C sapidus [117], the specific

ele-vation of MIH coincides with mid-vitellogenesis, when Vg

transcription and translation is intense, which is noteworthy

[117] The stimulatory role of MIH in Vg synthesis is

further substantiated by the specific binding to its receptors

in the hepatopancreas followed by the modulation of a

cAMP pathway involved in the Vg synthesis of C sapidus

In contrast, the mode of action of MIH on the Y-organs

occurs via binding to high-affinity receptors and increasing

the levels of cGMP in C maenas [118] Actinomycin D

blocks the stimulatory effects of MIH on Vg mRNA and

Vg synthesis, while cycloheximide lowers only Vg levels,

confirming the role of MIH in Vg transcription and

trans-lation [117] In this way, MIH and GIH have important

roles in the integrative control and coordination of molting

and reproduction in decapod crustaceans In the anecdysic

crab C sapidus, as well as the penaeid shrimp M ensis,

MIH achieves this coordination by stimulating Vg

synthesis and, at the same time, extending the intermolt

conditions, making it favorable for vitellogenic activities

Gonadotropin-releasing hormone

The recent discovery of the neuropeptide

gonadotropin-releasing hormone (GnRH) in the central nervous system

(CNS) of M rosenbergii gives further evidence that

verte-brate-like steroidal control of vitellogenesis is possible in

crustaceans [119] GnRH is a well-known decapeptide

ini-tiating hormonal induction in the brain–pituitary–gonadal

axis in vertebrates [120] Several studies have reported on

the occurrence of GnRH or GnRH-like peptides in fied invertebrate phyla ranging from corals to prochordates

diversi-In many invertebrates, especially the molluscs, the synthesis

of GnRH is related to reproductive activities [119] GnRHpeptides have been demonstrated by immunocytochemistry

in the CNS of the giant freshwater prawn M rosenbergii;however, interestingly enough, they are also found in theprevitellogenic as well as early vitellogenic oocytes, sug-gestive of a specific role in ovarian maturation in this prawn[119] Likewise, in the tiger prawn P monodon, GnRH-Iimmunoreactivity was also localized to the follicular cells ofproliferative, vitellogenic, and mature ovaries [121] Asexpected, ir-GnRH in shrimp was more closely related tooctGnRH and lGnRH-III than to other forms Hepatopan-creatic extract from P monodon could induce luteinizinghormone (LH) release from rat anterior pituitary glands invitro, demonstrating the potential role of LH in crustaceanreproductive function [122] In M rosenbergii and

P monodon, immunoreactivity for GnRH has also beenfound in the neurons as well as the nerve fibres innervatingsuch neurons in thoracic ganglia, suggesting that they reg-ulate the synthesis and release of serotonin, as well as ofGSH neuropeptides that are involved in the stimulation ofoocyte maturation Moreover, their occurrence in late pre-vitellogenic and early vitellogenic oocytes could imply astimulatory role for GnRH in the synthesis and release of sexsteroids in the ovaries of these decapods, as reported in theprotochordate Ciona intestinalis [123] Alternatively, theycould be involved in the control of ovarian maturation andovulation, as in mollusks [124] With this preliminary data, it

is not possible to postulate a mechanism for GnRH action ingonadal control in Crustacea, especially in the absence of averterbrate-like pituitary in these invertebrates However,

an early report is suggestive of the stimulation of oogenesis

in the sand shrimp Crangon crangon by a human tropin [80]

gonado-Methyl farnesoate and farnesoic acidHinsch [125] first revealed a functional role for MF from herobservations that the active MO implants stimulated ovariangrowth in the immature female spider crab Libinia emargi-nata Subsequent measurement of MF levels in the hemo-lymph and MOs demonstrated increased synthesis andsecretion of MF during vitellogenesis in this crab, suggest-ing a role in crustacean reproduction [126] In the blue crabCancer pagurus, MF concentrations in the hemolymphvaried throughout ovarian development, exhibiting a peak atthe beginning of secondary vitellogenesis (140 lg/ml) andfalling to basal levels thereafter [102] Rodriguez et al [127]showed the positive effects of MF on oocyte growth when

MF was injected alone or in combination with 17b-estradiol

in the crayfish Procambarus clarkii In addition, a higher

level of incorporation of labeled leucine was also induced by

MF in isolated pieces of ovary Obviously, MF has a positive

influence on vitellogenic activity within the ovary of this

crayfish Interestingly, crayfish ovary has been repeatedly

shown to engage in the autosynthesis of yolk proteins, unlike

many other decapod crustaceans [1] That MF has an

influ-ence on Vg uptake into the ovary was also revealed in the

finding that MF injection activated protein kinase C, an

isoenzyme involved in Vg uptake by oocytes and follicle

cells of the crayfish Cherax quadricarinatus [128]

However, recent studies have indicated that MF has no

effects on Vg gene expression in the hepatopancreas of

shrimp (M ensis), lobster (H americanus), and crab

(Cha-rybdis feriatus) [19,42,43] Using in vitro explant culture,

the above studies showed that the treatment of

hepatopan-creas fragments with FA, a precursor to MF, resulted in the

enhanced expression of the Vg gene In a more recent study

on H americanus, Tiu et al [129] revealed that the

com-bined use of FA and 20E increased HaVg1 gene expression

synergistically in the hepatopancreas

Previous studies indicated three major physiological

effects of MF: regulation of reproduction, molting [130–

132], and juvenile development [133] Hence, in many of

the reported results on reproduction, MF’s effect may result

from its action as a juvenilizing or molting hormone

Hence, the physiological status of the experimental animal

is a significant factor in deciding its effect However, in a

recent report on the shrimp Penaeus monodon, Marsden

et al [134] indicated an inhibitory role for MF in the late

stage of ovary development, which is unlikely to be due to

its molting or juvenilizing effect MF also has a purported

role in male reproduction and sexual behavior [135–137]

Nevertheless, recent gene expression studies (reviewed

above) may indicate a combinatorial effect of FA with 20E

on Vg synthesis in the hepatopancreas of several decapods

Crustaceans do not produce juvenile hormone (JH), but

several workers have used exogenous JHs to induce

vitel-logenesis in decapod crustaceans The effects of JH on

crustacean vitellogenesis vary greatly among species For

example, methoprene inhibited vitellogenesis in the

xant-hid crab Rhithropanopeus harrisii [138], but seemed to

promote vitellogenesis in the spider crab Libinia

emargi-nata [127] In the field crab Paratelphusa hydrodromous,

Sasikala and Subramoniam [139] found a stimulatory

effect of injected JH on vitellogenesis Despite the

non-occurrence of JH in crustaceans, a JH-responsive element

has been recently reported in the promoter region of the

Vg gene of a cladoceran, Daphnia magna [48] These

authors showed the occurrence of a nucleotide sequence,

in-between two Vg genes, that resembled juvenile

hormone-responsive and ecdysone-responsive elements in

D magna When JH agonists, such as pyriproxyfen and

fenoxycarb were injected, there was a strong repression of

the Vg gene expression in D magna The occurrence ofJH-responsive elements in Daphnia Vg may be only aremnant of the progressive evolution to other hormonaleffector molecules that developed later in crustaceans

Steroidal control of vitellogenesisEcdysteroids

Ecdysteroids are the principal hormonal factors in theinducement of molting in all arthropods They also play adefinitive role in the transcriptional activation of the Vggene in dipteran insects and certain ticks and mites [140].Although a similar role for ecdysteroids in crustaceanvitellogenesis is inconclusive, several reports implicate itsrole in female reproductive activity For example, Arvy

et al [141] found evidence that there is a rise in lymph ecdysteroids coincident with the initial stages ofgametogenesis, e.g., oogonial and spermatogonial mitoses

hemo-in the shore crab Carchemo-inus maenas In amphipods andisopods, hemolymph Vg levels parallel ecdysteroid titersduring the vitellogenic cycle, suggesting a role in Vgsynthesis [142] In the shrimp Lysmata seticaudate, vitel-logenin synthesis occurs under a high titer of ecdysone[143] Similarly, in the freshwater prawn M nipponense,Okumura et al [144] found a close correlation betweenhemolymph ecdysteroid titer and the corresponding ovar-ian maturation stages during the reproductive molt cycle.Similarly, in the crab E asiatica, the 20-hydroxy ecdysone(20E) titer in the hemolymph showed a gradual rise in theintermolt stage corresponding to ovarian maturation,whereas there was a rapid increase in 20E levels during thepremolt stage [71] In the lobster H americanus, 20E couldalso stimulate HaVg1 gene expression in the ovary alone or

in combination with FA [42] Notwithstanding these tive effects of ecdysteroids on vitellogenesis, Demeusy[145] showed the total noninvolvement of ecdysteroids invitellogenesis of the shore crab Carcinus maenas, asY-organ removal did not halt this process In the anecdysicoxyrhynchan crab Acanthonyx lunulatus, the Y-organdegenerates at the pubertal molt, and there are then twomore vitellogenic cycles that are completed in the absence

posi-of ecdysteroids [146] In M rosenbergii and P monodon,both hemolymph and ovarian ecdysteroid levels declinedfrom the immature to the late vitellogenic ovarian stages[147,148]

The question of ecdysteroid control of vitellogenin thesis in Crustacea can be resolved only by molecularstudies pertaining to their receptor activities In crustaceans,ecdysteroid receptor (EcR) has been identified in the blas-timal tissues of regenerating limbs of Uca pugilator, but itdimerizes with retinoid X receptor (RXR) [149] In another

syn-study, Durica et al [150] found the coexpression of these

two receptors (UpEcR and UpRXR) in the ovary of

U pugilator during the ovarian cycle, suggesting that the

ovary is a potential target tissue for ecdysteroid action

Further work is necessary to ascertain whether this receptor

activity is related to the ovarian synthesis of yolk protein

Interestingly, Tokishita et al [48] have recently described

the occurrence of an ecdysone-responsive element in the

upstream of the vitellogenin gene in a cladoceran, Daphnia

magna This, along with other binding sites for E74, E75,

and those for GATA factors in the D magna genome,

sug-gest that ecdysteroids activate the transcription of D magna

Vg genes, as in insects [151], although this activation is

antagonized by JH agonists [152] It should be noted here

that there is significant sequence homology between

D magna and insect Vgs (see above)

Vertebrate steroids

Vertebrate steroids such as estradiol and progesterone,

toge-ther with their metabolic products, have been identified in the

ovary and hepatopancreas of several decapod crustacean

species [78] These steroid hormones exhibit characteristic

fluctuations during gonadal maturation, indicating a role in

the control of reproduction In the tiger prawn Penaeus

monodon, both 17b-estradiol- and progesterone in free and

conjugated forms increase in the ovary during vitellogenesis,

but fall in the post-vitellogenic stages [153] Two of their

metabolic precursors, pregnenolone and

dehydroepiandros-terone, also show a peak during the major vitellogenic stages

within the ovary, suggesting that a biosynthetic pathway is

operational in the crustacean ovary, in a similar manner to that

of vertebrates In P monodon, 17b-estradiol and progesterone

levels in the hemolymph, hepatopancreas and ovary were also

shown to fluctuate closely with those of serum vitellogenin

levels during ovarian maturation [154] Similar fluctuations

of these hormones during the ovarian cycle have also been

reported in several decapods such as the brachyuran crab

S serrata [28], the spiny lobster P homarus [155], the

anomuran crab E asiatica [156], and the giant freshweater

prawn M rosenbergii [156] In M rosenbergii, such

hor-monal fluctuations were found only during the reproductive

molt cycle, whereas during the nonreproductive molt

char-acterized by a nondeveloping inactive ovary, the level of

estradiol in the hemolymph was not detectable at any molt

stage The immature ovary and hepatopancreas showed only

basal levels of estrogen during the nonreproductive molt

cycle, and progesterone levels were totally undetectable

These studies suggest that estradiol potentially plays a role in

crustacean vitellogenesis, either by upregulating Vg synthesis

as in vertebrates, or by stimulating certain metabolic

path-ways initiated during vitellogenesis, such as lipogenesis and/

or ion transport

Further evidence of the influence of vertebrate steroids onvitellogenesis has been adduced from the injection ofexogenous hormones In Penaeus japonicus, injection ofprogesterone and 17a-hydroxyprogesterone induced ovarianmaturation in M ensis [157] and stimulated Vg secretion in

P japonicus [158] In vitro culture of previtellogenic ovary

of immature M japonicus with 17b-estradiol resulted in theinducement of Vg synthesis into the medium, as well as theappearance of primary vitellogenic oocytes [159] Simi-larly, explants of hepatopancreas of M ensis, incubated invitro with 17b-estradiol and progesterone stimulated VgmRNA synthesis in the early vitellogenic ovary of thecrayfish Cherax albidus [160] Thus, 17b-estradiol andprogesterone have positive effecte on Vg synthesis both inthe ovary and hepatopancreas in these species In the am-phipods, a vitellogenesis-stimulating ovarian hormone(VSOH) has been proposed to induce Vg synthesis in the fatbody [161] With the knowledge that the ovary is the site ofsynthesis of the vertebrate sex steroids that stimulate Vgsynthesis, could these steroids then be the VSOH suggestedfor amphipods?

Injection of exogenous estradiol resulted in the tion of vitellogenesis in the ovary of premature female sandcrab E asiatica, in addition to eliciting a new Vg fraction inthe hemolymph Significantly, M rosenbergii in thenonreproductive molt cycle also initiated vitellogenesisfollowing estradiol injection (unpublished observation)

stimula-In oviparous vertebrates, the synthesis of vitellogenin istighhtly controlled by an estrogen hormone signal trans-duction pathway, which is mediated by estrogen receptorand heat shock protein 90 (Hsp90) [162] A recent report by

Wu and Chu [163] on the Hsp90 activity during genesis in M ensis has indicated a strong correlationbetween estrogen hormones and Hsp90 expression,suggesting that the expression of Vg may be under theregulation of estrogen through a mechanism similar to that invertebrates

vitello-The occurrence of nuclear receptors for both terone (PR) and estrogen (ER) has recently been reported

proges-in the freshwater crayfish Austropotamobius pallipes [164]

By using immunohistochemistry and western blottingapproaches, these authors showed the presence of PR in theovary and hepatopancreas and ER only in the hepatopan-creas of this crayfish Whereas ER has a direct role in thetranscriptional control of Vg gene in hepatopancreas, thepresence of PR in the hepatopancreas of A pallipes sug-gests that progesterone plays a genomic role mediated byits receptor, as opposed to the hypothesized function ofprogesterone as a precursor of estradiol necessary forvitellogenesis [165] EST analysis of the cDNA libraryestablished from vitellogenic ovary of P mondon revealedthe expression of a progesterone receptor-releated proteinP23 (Pm-p23) during vitellogenesis In situ hybridization

also indicated that Pm-p23 was localized in the ooplasm of

previtellogenic oocytes [166]

Future perspectives

A proper understanding of all aspects of vitellogenesis is

necessary to formulate control measures to increase egg

production in commercially important decapod

crusta-ceans Recent molecular and immunological approaches to

the study of crustacean vitellogenesis have appreciably

improved our understanding of the mechanism and control

of yolk formation Gene expression studies relating to

vitellogenin synthesis in different organs including the

ovary have shed new light on resolving the problem of yolk

precursor synthetic sites A general consensus that has

emerged from gene expression studies is that penaeid

shrimp produce yolk both in ovary and hepatopancreas,

whereas in many other decapods such as crabs and lobsters,

vitellogenin synthesis is restricted to the hepatopancreas

Yolk synthesis within the ovary may be considered anearlier evolutionary feature in marine shrimp in view ofscanty yolk in the egg combined with the reproductivebehavior of free spawning and the occurrence of naupliuslarvae Conversely, crabs and other species belonging tothe suborder Pleocyemata lay numerous yolk-laden eggsand incubate them in a brood until the eggs hatch intoadvanced zoea larvae These forms naturally rely on thehepatopancreas for increased yolk precursor production.Our knowledge of the hormonal control of vitellogenesis

is mainly based on experimental studies involving theextirpation or implantation of endocrine organs of decapodcrustaceans With the advent of gene expression studies on

Vg synthesis using in vitro culture systems, verification ofhormonal influence at the transcriptional control levelbecomes possible In this respect, the cross-functionalactivities of the CHH family of peptides have revealed theirmultiple controlling effects on both vitellogenesis andmolting Furthermore, hormone receptor expression studiesduring vitellogenesis have also adduced evidence on

Fig 6 A hypothetical model of the neuroendocrine control of

vitellogenesis in decapod crustaceans Different endocrine pathways

included in this diagram reveal the involvement of various hormonal

factors operating in different crustacean species (see main text for

details) The existence of inhibitory peptides may be unique to

crustaceans, but their cross-functional activity likely fine-tunes the

regulation of vitellogenic activities PVO previtellogenic ovary,

Vg vitellogenin, MF methyl farnesoate, FA farnesoic acid, 20E 20-hydroxyecdysone, VSOH vitellogenesis-stimulating ovarian hor- mone, GnRH gonadotropin-releasing hormone Solid line stimulatory, broken line inhibitory

proximate endocrine factors controlling yolk precursor

synthesis in several crustacean species Vitellogenin gene

expression occurring under the specific influence of FA

secreted by the MOs and 20E in the hepatopancreas as well

as the ovary provides direct evidence that these two organs

are involved in yolk synthesis in shrimp species and to a

certain extent lobsters Analysis and identification of the

hormone-responsive elements occuring in the Vg gene

regulatory regions are expected to throw more light on this

aspect The identification of DNA sequences comparable to

these insect hormone-responsive elements in the brachiopod

Daphnia not only points to this possibility, but also indicates

the relatedness of the hormonal regulation of vitellogenesis

between insects and crustaceans In addition, vertebrate sex

steroids have been thought to have a role in the control of

vitellogenesis in as much as the same hormones control egg

maturation in vertebrates Expression profiles of estrogen

receptor and progestrone receptor in reproductive tissues

during vitellogenesis in the freshwater crayfish lend support

to the transcriptional control of Vg synthesis by estrogen in

these crustaceans Evidently, crustaceans employ multiple

hormonal factors often synergistically in the control of

vitellogenesis and related reproductive phenomena in order

to successfully accomplish egg production without affecting

somatic growth Figure6depicts the hypothetical hormonal

controlling mechanisms operating in typical decapod

crus-taceans The occurrence of multiple hormonal factors in

crustaceans to control reproduction obviously arises from

the condition that both molting and reproduction occur

either sequentially or in an overlapping manner in various

species Nonetheless, the evolution of endocrine regulatory

mechanisms to control Vg gene expression appears to occur

in the insectan line It seems promising to imagine that the

common ancestor of insects and crustaceans possessed all

these hormonal factors Whereas insects went on to use JH

and 20E as the chief gonadotropic hormones in egg

pro-duction, crustaceans experimented with a variety of

hor-monal effector molecules, probably in an attempt to

augment yolk production to meet the enormous demands of

embryonic nutrition Incidentally, the use of steroidal

hor-mones in Vg gene regulation by vertebrates would have

originated from invertebrates With all the recent

advance-ments in understanding the endocrine control of crustacean

reproduction, it is now time for the reproductive

endocri-nologists to use such information in their efforts to control

the reproduction of commercially significant crustaceans

Acknowledgments This review is fondly dedicated to Dr Marcy

N Wilder of the Japan International Research Center for Agricultural

Sciences (JIRCAS), Tsukuba, whose scientific association has

sig-nificantly enhanced my understanding of crustacean vitellogenesis I

am thankful to my students, Dr Vidya Jayasankar of the Central

Marine Fisheries Research Institute, Chennai, and Dr Sudha Warrier

of the Manipal Medical College, Bangalore, for critically reading

through the manuscript I also thank Dr C.P Balasubramanian and

Dr Sherly Tom of the Central Institute of Brackish Water ture, Chennai for useful discussion I am also grateful to Mr S Muthu Kumar of the National Institute of Ocean Technology, Chennai for his help in the homology study of vitellogenins I finally thank the Indian National Science Academy for the award of the position of Senior Scientist.

Aquacul-References

1 Adiyodi RG, Subramoniam T (1983) Arthropoda—Crustacea In: Adiyodi KG, Adiyodi RG (eds) Reproductive biology of invertebrates, vol I Oogenesis, oviposition, oosorption Wiley, New York, pp 443–495

2 Wilder MN, Subramoniam T, Aida K (2002) Yolk proteins of Crustacea In: Raikhel AS, Sappington TW (eds) Reproductive biology of invertebrates, vol XII (part A) Science Publishers Inc., Enfield, pp 131–174

3 Wallace RA, Walker SL, Hausehka PV (1967) Crustacean lipovitellin Isolation and characterization of the major high- density lipoprotein from the eggs of decapods Biochemistry 6:1582–1590

4 Tirumalai R, Subramoniam T (1992) Purification and terization of vitellogenin and liprovitellins of the sand crab Emerita asiatica: molecular aspects of crab yolk proteins Mol Reprod Dev 33:16–26

charac-5 Thirumalai R, Subramoniam T (2001) Carbohydrate nents of lipovitellin of the sand crab Emerita asiatica Mol Reprod Dev 58:54–62

compo-6 Khalaila I, Peter-Katalinic J, Tsang C, Radcliffe M, Aflahu D, Harvey DJ, Dwek RA, Rudd PM, Sagi A (2004) Structural characterization of the N-glycan moity and site of glycosylation

in vitellogenin from the decapod crustacean Cherax rinatus Glycobiology 14:767–774

quadrica-7 Meusy JJ, Payen G (1988) Female reproduction in can Crustacea Zool Sci 5:217–265

malacostra-8 Lui CW, O’ Connor JD (1977) Biosynthesis of crustacen lipovitellin III The incorporation of labeled amino acids into the purified lipovitellin of the crab Pachygrapsus crassipes J Exp Zool 195:105–108

9 Yano I, Chinzei Y (1987) Ovary is the site of vitellogenin synthesis in kuruma prawn, Penaeus japonicus Comp Biochem Physiol 86B:213–218

10 Fainzilber M, Tom M, Shafir S, Applebaum SW, Lubzens E (1992) Is there extraovarian synthesis of vitellogenin in penaeid shrimp? Biol Bull 183:233–241

11 Kim YK, Tsutsui N, Kawazoe I, Okumura T, Kaneko T, Aida K (2005) Localization and developmental expression of mRNA for cortical rod protein in kuruma prawn Marsupenaeus japonicus Zool Sci 22:675–680

12 Okumura T, Kim YK, Kawazoe I, Yamano K, Tsutsui N, Aida K (2006) Expression of vitellogenin and cortical rod proteins during induced ovarian development by eyestalk ablation in the kuruma prawn, Marsupenaeus japonicus Comp Biochem Physiol 143A:246–253

13 Khayat M, Lubzens E, Tietz A, Funkenstein B (1994) Are vitellin and vitellogenin coded by one gene in the marine shrimp Penaeus semisulcatus? J Mol Endocrinol 12:251–254

14 Tsutsui N, Kawazoe I, Ohira T, Jasmani S, Yang W-J, Wilder MN, Aida K (2000) Molecular characterization of a cDNA encoding vitellogenin and its expression in the kuruma prawn, Penaeus japonicus Zool Sci 17:651–660

15 Tsang WS, Quackenbush LS, Chow BK, Tiu SH, He JG, Chan SM (2003) Organization of the shrimp vitellogenin gene Evidence of

multiple genes and tissue specific expression by the ovary and

hepatopancreas Gene 303:99–109

16 Tsutsui N, Katayama H, Ohira T, Nagasawa H, Wilder MN,

Aida K (2005) The effects of crustacean hyperglycemic

hor-mone-family peptides on vitellogenin gene expression in the

kuruma prawn, Marsupenaeus japonicus Gen Comp Endocrinol

144:232–239

17 Avarre J-C, Michelis R, Tietz A, Lubzens E (2003) Relationship

between vitellogenin and vitellin in a marine shrimp (Penaeus

semisulcatus) and molecular characterization of vitellogenin

complementary DNAs Biol Reprod 69:355–364

18 Phiriyangkul P, Puengyam P, Jakobsen IB, Utarabhand P (2007)

Dynamics of vitellogenin mRNA expression during

vitellogen-esis in the banana shrimp Penaeus (Fenneropenaeus

merguien-sis) using real-time PCR Mol Reprod Dev 74:1198–1207

19 Kung SY, Chan SM, Hui JH, Tsang WS, Mak A, He JG (2004)

Vitellogenesis in the sand shrimp, Metapenaeus ensis: the

con-tribution from the hepatopancreas-specific vitellogenin gene

(MeVg2) Biol Reprod 71:863–870

20 Tiu SH, Hui JH, Mak AS, He JG, Chan SM (2006) Equal

contribution of hepatopancreas and ovary to the production of

vitellogenin (PmVg 1) transcripts in the tiger shrimp, Penaeus

monodon Aquaculture 254:666–674

21 Lee RF, Puppione DL (1998) Lipoproteins I and II from the

hemolymph of the blue crab Callinectes sapidus: lipoprotein II

associated with vitellogenesis J Exp Zool 248:278–279

22 Komatsu M, Andi S (1998) A very-high-density lipoprotein with

clotting ability from hemolymph of sand crayfish, Ibcacus

cili-atus Biosci Biotechnol Biochem 62:459–463

23 Lubzens E, Ravid T, Khayat M, Daube N, Tiez A (1997)

Iso-lation and characterization of the high-density lipoproteins from

the hemolymph and ovary of the penaeid shrimp Penaeus

semisulcatus (de Haan): apoproteins and lipids J Exp Zool

278:339–348

24 Yehezkel G, Chayoth R, Abdu U, Khalaila I, Sagi A (2000)

High-density lipoprotein associated with secondary

vitellogen-esis in the hemolymph of the crayfish Cherax quadricarinatus.

Comp Biochem Physiol B 127:411–421

25 Subramoniam T, Gunamalai V (2003) Breeding biology of the

sand crab Emerita asiatica (Decapoda: Anomura) Adv Mar

Biol 46:91–182

26 Subramoniam T, Tirumalai R, Gunamalai V, Hoffmann KH

(1999) Embryonic ecdysteroids in a mole crab Emerita asiatica

(Miline Edwards) J Biosci 24:91–96

27 Warrier S, Tirumalai R, Subramoniam T (2001) Occurrence of

vertebrate steroids, estradiol 17b and progesterone in the

reproducing females of the mud crab Scylla serrata Comp

Biochem Physiol 130:283–294

28 Kerr MS (1968) Protein synthesis by hemocytes of Callinectes

sapidus: a study of in vitro incorporation of14C leucin J Cell

Biol 39:72a–73a

29 Ezhilarasi S, Subramoniam T (1986) Serological studies on the

egg maturation in the edible crab, Scylla serrata J Singap Natl

Acad Sci 15:21–25

30 Suzuki S, Yamasaki K, Katakura Y (1990) Vitellogenin synthesis

in andrectomized males of the terrestrial isopod, Armadillidium

vulgare (Malacostracan Crusacea) Biol Bull 77:120–126

31 Junera H, Croisille Y (1980) Recherche du lieu de synthese de la

vitellogenine chez le Crustace Amphipode Orchestia gammarella

(Pallas) Mise en evidence d’une activation de is synthese

pro-teique dans le tissue adipeux sous-epidermique en liaison avec la

production de vitellogenine C R Acad Sci Paris 290:703–706

32 Meusy JJ, Junera H, Cledon P, Martin M (1983) La

vitelloge-nine chez un Crustace Decapod Natantia Palaemon serratus

Pennant Mise en evidence comparaison immunologique avec

les viellines, site de synthese et role des pedoncules oculaires Reprod Nutr Dev 23:625–640

33 Rani K, Subramoniam T (1997) Vitellogenesis in the mud crab Scylla serrata—an in vivo isotope study J Crustac Biol 17:659–665

34 Chen YN, Tseng DY, Ho PY, Kuo CM (1999) Site of genin synthesis determined from a cDNA encoding a vitello- genin fragment in the freshwater giant prawn Macrobrachium rosenbergii Mol Reprod Dev 54:215–222

vitello-35 Yang W-J, Ohira T, Tsutsui N, Subramoniam T, Huong DTT, Aida K, Wilder MN (2000) Determination of amino acid sequence and site of expression of four vitellins in the giant freshwater prawn, Macrobrachium rosenbergii J Exp Zool 287:413–422

36 Jayasankar V, Tsutsui N, Jasmani S, Saido-Sakanaka H, Yang W-J, Okuno A, Tran TT, Aida K, Wilder MN (2002) Dynamics of vitellogenin mRNA expression and changes in hemolymph vitel- logenin levels during ovarian maturation in the giant freshwater prawn Macrobrachium rosenbergii J Exp Zool 293:675–682

37 Khayat M, Lubzens E, Tietz A, Funkenstein B (1994) Cell-free synthesis of vitellin in the shrimp Penaeus semisulcatus (de Haan) Gen Comp Endocrinol 93:205–213

38 Li K, Chen L, Zhou Z, Li E, Zhao X, Guo H (2006) The site of vitellogenin synthesis in Chinese mitten-handed crab Eriocheir sinensis Comp Biochem Physiol B 143:453–458

39 Chan SM, Mak AS, Choi CL, Ma TH, Hui JH, Tiu SH (2005) Vitellogenesis in the red crab Charybdis feriatus Contributions from small vitellogenin transcripts (CfVg) and farnesoic acid stimulation of CfVg expression Ann NY Acad Sci 1040:74–79

40 Warrier S, Subramoniam T (2003) Instability of crab genin and its immunological relatedness with mammalian ath- erogenic lipoproteins Mol Reprod Dev 64:329–340

vitello-41 Zmora N, Trant J, Chan SM, Chung JS (2007) Vitellogenin and its messenger RNA during ovarian development in the female blue crab Callinectes sapidus: gene expression, synthesis, transport, and cleavage Biol Reprod 77:138–146

42 Tiu SH, Hui HL, Tsukimura B, Tobe SS, He JG, Chan SM (2009) Cloning and expression study of the lobster (Homarus americanus) vitellogenin: conservation in gene structure among decapods Gen Comp Endocrinol 160:36–46

43 Mak ASC, Choi CL, Tiu SHK, Hui JHG, Tobe SS, Chan S (2005) Vitellogenesis in the red crab Charybdis feriatus: hepa- topancreas specific expression and farnesoic acid stimulation of vitellogenin gene expression Mol Reprod Dev 7:288–300

44 Babin PJ, Bogerd J, Kooiman FP, Van Marrewijk WJA, Van der Horst DJ (1999) Apolipophorin II/I, apolipoprotein B, vitello- genin, and microsomal triglyceride transfer protein genes are derived from a common ancestor J Mol Evol 49:150–160

45 Sundermeyer K, Hendricks JK, Prasad SV, Wells MA (1996) The precursor protein of the structural apolipoproteins of lipo- phorin: cDNA and deduced amino acid sequence Insect Bio- chem Mol Biol 26:735–738

46 Voloch CM, Freire PR, Russo CA (2005) Molecular phylogeny

of penaeid shrimps inferred from two mitochondrial markers Genet Mol Res 4:668–674

47 Kato Y, Tokishita SI, Ohta T, Yamagata H (2004) A genin chain containing a superoxide dismutase-like domain is the major component of yolk proteins in cladoceran crustacean Daphnia magna Gene 334:157–165

vitello-48 Tokishita S, Kato Y, Kobayashi T, Nakamura S, Ohta T, Yamagata H (2006) Organization and repression by juvenile hormone of a vitellogenin gene cluster in the crustacean, Daphnia magna Biochem Biophys Res Commun 345:362–370

49 Matsumoto T, Nakamura AM, Mori K, Kayano T (2003) Molecular characterization of a cDNA encoding putative vitellogenin from the Pacific oyster Crassostrea gigans Zool Sci 20:37–42

50 Hayakawa H, Andoh T, Watanabe T (2006) Precursor structure

of egg proteins in the coral Galaxea fascicularis Biochem

Biophys Res Commun 344:173–180

51 Avarre J-C, Lubzen E, Babi PJ (2007) Apolipocrustacein, formerly

vitellogenin, is the major egg yolk precursor protein in decapod

crustaceans and is homologous to insect apolipophorin II/I and

vertebrate apolipoprotein B BMC Evol Biol 7:3–13

52 Baker ME (1988) Is vitellogenin an ancestor of apolipoprotein

B-100 of human lipoprotein lipase? Biochem J 255:1057–1060

53 Jugan P, Soyez D (1985) Demonstration in vitro del’ inhibition

de I’inhibition de I’ endocytose ovocytaire par un extrait de

glande de sinus chez la crevette Macrobrachium rosenbergii.

C R Acad Sci Paris Ser III 300:705–709

54 Laverdure A-M, Soyez D (1988) Vitellogenin receptor from

lobster oocyte membrane: solubilization and characterization by a

solid phase binding assay Int J Invertebr Reprod Dev 13:251–266

55 Jugan P, Van Herp F (1989) Introductory study of an oocyte

membrane protein that specifically binds vitellogenin in the

crayfish, Orconectus limosus Invertebr Reprod Dev 16:149–154

56 Warrier S, Subramoniam T (2002) Receptor mediated yolk

protein uptake in the crab Scylla serrata: crustacean vitellogenin

receptor recognizes related mammalian serum lipoproteins Mol

Reprod Dev 61:536–548

57 Tiu SH, Benzie J, Chan SM (2008) From hepatopancreas to ovary:

molecular characterization of a shrimp vitellogenin receptor

involved in the processing of vitellogenin Biol Reprod 79:66–74

58 Mekuchi M, Ohira T, Kawazoe I, Jasmani S, Suitoh K, Kim YK,

Jayasankar V, Nagasawa H, Wilder MN (2008) Characterization

and expression of the putative ovarian lipoprotein receptor in the

kuruma prawn, Marsupenaeus japonicus Zool Sci 25:428–437

59 Schneider WJ (1992) Vitellogenin receptors: oocyte-specific

members of the low-density lipoprotein receptor superfamily.

Int Rev Cytol 166:103–137

60 Atella GC, Silva-Neto MAC, Golodne DM, Arifin S, Shahabuddin

M (2006) Anopheles gambusia lipophorin characterization and

role in lipid transport to developing oocytes Insect Biochem Mol

Biol 36:375–386

61 Raikhel AS, Dhadialla TS (1992) Accumulation of yolk proteins

in insect oocytes Ann Rev Entomol 37:217–251

62 Byrne BM, Gruber M, Ab G (1989) The evolution of yolk

proteins Prog Biophys Mol Biol 53:33–69

63 Suzuki S (1987) Vitellins and vitellogenins of the terrestrial

isopod Armadillidium vulgare Biol Bull 173:345–357

64 Okuno A, Katayama H, Nagasawa H (2000) Partial characterization

of vitellin and localization production in the terrestrial isopod,

Armadillidium vulgare Comp Biochem Physiol 126B:397–407

65 Okuno A, Yang W-J, Jayasankar V, Saido-Sakanaka H,

Huong DTT, Jasmani S, Atmomarsono M, Subramoniam T,

Tsutsui N, Ohira T, Kawazoe I, Aida K, Wilder MN (2002)

Deduced primary structure of vitellogenin in the gaint

fresh-water prawn, Macrobrachium rosenbergii and yolk processing

during ovarian maturation J Exp Zool 292:417–429

66 Raviv S, Parnes S, Segall C, Sagi A (2006) Complete sequence

of Litopenaeus vannamei (Crustacea: Decapod) vitellogenin

cDNA and its expression in endocrinologically induced

sub-adult females Gen Comp Endocrinol 145:39–50

67 Kang BJ, Nantri T, Lee JM, Saito H, Han CH, Hatakeyama M,

Saigusa M (2008) Vitellogenesis in both sexes of gonochoristic

mud shrimp, Upogebia major (Crustacea): analyses of

vitello-genin gene expression and vitellovitello-genin processing Comp

Bio-chem Physiol B 149:589–598

68 Komatsu M, Ando S (1992) A novel low-density lipoprotein with

large amounts of phospholipid found in the egg yolk of sand

crayfish Ibacus ciliatus: its function as vitellogenin-degrading

proteinase Biochem Biophys Res Commun 186:498–502

69 Subramoniam T (2007) Embryonic nutrition and yolk utilization

in the sand crab Emerita asiatica J Endocrinol Reprod 11:1–14

70 Subramoniam T (2000) Crustacean ecdysteroids in reproductin and embryogenesis Comp Biochem Physiol 125C:135–156

71 Gunamalai V, Kirubagaran R, Subramoniam T (2004) Hormonal coordination of molting and female reproduction by ecdyster- oids in the mole crab Emerita asiatica (Milne Edwards) Gen Comp Endocrinol 138:128–138

72 Walker A, Ando S, Smith GD, Lee RF (2006) The utilization of lipovitellin during blue crab (Callinectes sapidus) embryogen- esis Comp Biochem Physiol B Biochem Mol Biol 143:201–208

73 Shimitzu K, Satuito CG, Saikawa W, Fusetani W (1996) Larval storage protein of the barnacle, Balanus amphitrite: biochemical and immunological similarities to vitellin J Exp Zool 276:87–94

74 Dreanno C, Matsumura K, Dohmae N, Takio K, Hirota H, Kirby

RR, Clare AS (2006) An (alpha) 2-macroglobulin-like protein is the cue to gregarious settlement of the barnacle Balanus am- phitrite Proc Natl Acad Sci USA 103:14396–14401

75 Subramoniam T (2004) Hormonal controls of female duction and molting in decapod crustaceans J Endocrinol Reprod 44:1–12

repro-76 Soyez D, Van Deijnen JE, Martin M (1987) Isolation and characterization of a vitellogenesis-inhibiting factor from sinus glands of the lobster, Homarus americanus J Exp Zool 244:479–484

77 Soyez D, Le Caer JP, Noel PY, Rossier J (1991) Primary structure

of two isoforms of the vitellogenesis inhibiting hormone from the lobster Homarus americanus Neuropeptides 20:25–32

78 Subramoniam T (1999) Endocrine regulation of egg production

in economically important crustaceans Curr Sci 76:350–360

79 Edomi P, Azzoni E, Mettulio R, Pandolfelli N, Ferrero EA, Giulianini PB (2002) Gonad-inhibiting hormone of the Norway lobster (Nephrops norvegicus) cDNA cloning, expression, recombinant protein production, and immunolocalization Gene 284:93–102

80 Bomirski A, Kelk-Kawinska E (1976) Stimulation of oogenesis

in the sand shrimp, Crangon crangon by a human gonadotropin Gen Comp Endocrinol 30:239–242

81 Quackenbush LS, Keeley LL (1988) Regulation of esis in the fiddler crab Uca pugilator Biol Bull 175:321–331

vitellogen-82 Browdy CL, Fainzilber M, Tom M, Loya L, Lubzens E (1990) Vitellin synthesis in relation to oogenesis in in-vitro incu- bated ovaries of Penaeus semisulcatus (Crustacea, Decapoda, Penaeidae) J Exp Zool 255:205–215

83 Aguilar MB, Quackenbush LS, Hunt DT, Shabanowitz J, Huberman A (1992) Identification, purification and initial characterization of the vitellogenesis-inhibiting hormone from the Mexican crayfish Procambarus bouvieri (Ortmann) Comp Biochem Physiol 102B:491–498

84 Vincent SGP, Keller R, Subramoniam T (2001) Development of vitellogenin-ELISA an in vivo bioassay and identification of two vitellogenesis-inhibiting hormones of the tiger shrimp, Penaeus monodon Mar Biotech 3:561–571

85 De Kleijn DPV, Janssen KPC, Waddy SL, Hegeman R, Lai WY, Martens GJM, Van Herp F (1998) Expression of the crustacean hyperglycemic hormones and the gonad-inhibiting hormone during the reproductive cycle of the female American lobster Homarus americanus J Endocrinol 156:291–298

86 Treerattrakool S, Panyim S, Chan SM, Withyachumnarnkul B, Udomkit A (2008) Molecular characterization of gonad-inhib- iting hormone of Penaeus monodon and elucidation of its inhibitory role in vitellogenin expression by RNA interference FEBS J 275:970–980

87 Tsutsui N, Ohira T, Kawazoe I, Takahashi A, Wilder MN (2007) Purification of sinus gland peptides having vitellogenesis-inhibiting

activity from the whiteleg shrimp Litopenaeus vannamei Mar

Biotech 9:360–369

88 Marco HG, Avarre JC, Lubzens E, Gade G (2002) In search of a

vitellogenesis-inhibiting hormone from the eyestalk of South

African spiny lobster, Jasus lalandi Invert Reprod Dev 41:

15–143

89 Ohira T, Okumura T, Suzuki M, Yajima Y, Tsutsui N, Wilder

NM, Nagasawa H (2006) Production and characterization of

recombinant vitellogenesis inhibiting hormone from the

Amer-ican lobster, Homarus amerAmer-icanus Peptides 27:1251–1258

90 Okumura T, Yamano K, Sakiyama K (2007) Vitellogenesis gene

expression and hemolymph vitellogenin during vitellogenesis,

final maturation and oviposition in female kuruma prawn,

Marsupenaeus japonicus Comp Biochem Physiol A 147:

1028–1037

91 Okumura T (2007) Effects of bilateral and unilateral eyestalk

ablation on vitellogenin synthesis in immature female kuruma

prawn Marsupenaeus japonicus Zool Sci 24:233–240

92 Jayasankar V, Jasmani S, Tsutsui N, Aida K, Wilder MN (2006)

Dynamics of vitellogenin synthesis in juvenile giant freshwater

prawn Macrobrachium rosenbergii J Exp Zool A Comp Exp

Biol 305:440–448

93 Greve P, Sorokine O, Berges T, Lacombe C, Van Dorsselaer A,

Martin G (1999) Isolation and amino acid sequence of a peptide

with vitellogenesis inhibiting activity from the terrestrial isopod

Armadillium vulgare (Crustacea) Gen Comp Endocrinol

115:406–414

94 Bowman CJ, Kroll KJ, Gross TG, Denslow ND (2002)

Estra-diol-17b induced gene expression in largemouth bass

(Micr-opterus salmoides) Mol Cell Endocrinol 196:67–77

95 Gohar M, Souty C (1983) Mise en e’vidence in vitro d’une

synthe et d’une libe’ration de vitelloge’nine dans le tissue

adi-peux male de Porcellio dilatatus, (Brandt) C R Acad Sci Paris

297:145–148

96 Tsutsui N, Saido-Sakanaka H, Yang W-J, Jayasankar V, Jasmani

S, Okuno A, Ohira T, Okumura T, Aida K, Wilder MN (2004)

Molecular characterization of a cDNA encoding vitellogenin in

the coonstriped shrimp, Pandalus hypsinotus and site of

vitel-logenin mRNA expression J Exp Zool A Comp Exp Biol

301:802–814

97 Abdu U, Davis C, Khalaila SagiA (2002) The vitellogenin

cDNA of Cherax quadricarinatus encodes a lipoprotein with

calcium binding ability, and its expression is induced following

the removal of the androgenic gland in a sexual plastic system.

Gen Comp Endocrinol 127:263–272

98 Sagi A, Manor R, Segall C, Davis C, Halaila I (2002) On

inter-sexuality in the crayfish Cherax quadricarinatus: an inducible

sexual plasticity model Invertebr Reprod Dev 41:27–33

99 Laufer H, Landau M, Homola E, Borst EW (1987) Methyl

farnesoate: its site of synthesis and regulation of secretion in a

juvenile crustacean Insect Biochem 17:1129–1131

100 Liu L, Laufer H (1996) Isolation and characterization of sinus

gland neuropeptides with both mandibular organ inhibiting and

hyperglycemic effects from the spider crab Libinia emarginata.

Arch Insect Biochem Physiol 32:375–385

101 Wainwrite G, Webster SG, Wilkinson MC, Chung JS, Rees HH

(1996) Structure and significance of mandibular organ-inhibiting

hormone in the crab, Cancer pagurus J Biol Chem 271:

12749–12754

102 Wainwrite G, Prescott MC, Rees HH, Webster SG (1996) Mass

spectrometric determination of methyl farnesoate profiles and

correlation with ovarian development in the edible crab Cancer

pagurus J Mass Spectrum 31:1338–1344

103 Wainwrite G, Websters SG, Rees HH (1998) Neuropeptide

regulation of biosynthesis of the juvenoid, methyl farnesoate in

the edible crab, Cancer pagurus Biochem J 334:651–657

104 Otsu T (1963) Bihormonal control of sexual cycle in freshwater crab, Potamon dehaani Embryologia 8:1–20

105 Hinsch GW, Bennett DC (1979) Vitellogenesis stimulated by thoracic ganglion implants into destalked immature spider crabs, Libinia emarginata Tissue Cell 11:345–351

106 Takayanagi H, Yamamoto Y, Takeda N (1986) An ovary stimulating factor in the shrimp Paratya compressa J Exp Zool 240:203–209

107 Yano I (1988) Hormonal control of vitellogenesis in penaeid shrimp In: Flegel TW (ed) Advances in shrimp biotechnology National Centre for Genetic Engineering and Biotechnology, Bangkok, pp 29–31

108 Richardson HG, Deecaraman M, Fingerman M (1991) The effects of biogenic amine on ovarian development in the fiddler crab Uca pugilator Comp Biochem Physiol 99C:53–247

109 Sarojini R, Nagabushanam R, Fingerman M (1996) In vitro inhibition by DA of 5-hydroxytryptamine stimulated ovarian maturation in the red swamp crayfish Procambarus clarkii Experientia 52:707–709

110 Sarojini R, Nagabushanam R, Fingerman M (1997) An in vitro study of the inhibitory action of methionine enkephalin on ovarian maturation in the red swamp crayfish Procambarus clarkii Comp Biochem Physiol 117C:207–210

111 Subramoniam T, Kirubagaran K (2010) Endocrine regulation of vitellogenesis in lobsters J Mar Biol Assoc India (in press)

112 Chen YN, Fan HF, Hsieh SL, Kuo CM (2003) Physiological involvement of DA in ovarian development of the freshwater giant prawn, Macrobrachium rosenbergii Aquaculture 228:383–395

113 De Kleijn DPV, Janssen KPC, Van Den Berg MC, Martens GJM, Van Herp F (1995) Cloning and expression of two mRNAs encoding structurally different crustacean hyperglyce- mic hormone precursors in the lobster Homarus americanus Biochim Biophys Acta 1260:62–66

114 Gu PL, Tobe SS, Chow BKC, Chu KH, He J-G, Chan S-M (2002) Characterization of an additional molt inhibiting hor- mone-like neuropeptide from the shrimp Metapenaeus ensis Peptides 23:1875–1883

115 Tiu SH, Chan SM (2007) The use of recombinant protein and RNA interference approaches to study the reproductive func- tions of a gonad-stimulating hormone from the shrimp Metap- enaeus ensis FEBS J 274:4385–4395

116 Zmora N, Trant J, Zohar Y, Chung JS (2009) Molt-inhibiting hormone stimulates advanced ovarian developmental stages in the female blue crab, Callinectes sapidus I: an ovarian stage dependent involvement Saline Syst 5:7

117 Zmora N, Sagi A, Zohar Y, Chung JS (2009) Molt-inhibiting hormone stimulates advanced ovarian developmental stages in the female blue crab, Callinectes sapidus 2: novel specific binding sites in hepatopancreas and cAMP as a second mes- senger Saline Syst 5:6

118 Chung JS, Webster SG (2006) Binding sites of crustacean hyperglycemic hormone and its second messengers on gills and hindgut of the green shore crab Carcinus maenus A possible osmoregulatory role Gen Comp Endocrinal 147:206–213

119 Ngernsoungnern A, Ngernsoungnern P, Kavanaugh S, Sower SA, Sobhon P, Sretarugsa P (2008) The identification and distribution

of gonadotropin releasing hormone-like peptides in the central nervous system and ovary of the giant freshwater prawn, Mac- robrachium rosenbergii Invertebr Neurosci 8:49–57

120 Millar RP (2005) GnRHs and GnRH receptors Anim Reprod Sci 88:5–28

121 Ngernsoungnern P, Ngernsoungnern A, Kavanaugh S, Sobhon

P, Sower SA, Sretarugsa P (2008) The presence and distribution

of gonadotropin releasing hormone-like factor in the central nervous system of the black tiger shrimp, Penaeus monodon Gen Comp Endocrinol 155:613–622

122 Fann MC, Man WC, Wang PS (1990) Existence of a

gonado-tropin-releasing hormone like factor in brass shrimp (black tiger

prawn, Penaeus monodon) Chin J Physiol 33:169–178

123 Di Fiore MM, Rastogi RK, Ceciliani F, Messi E, Botte V, Botte L,

Pinelli C, D’Aniello B, D’Aniello A (2000) Mammalian and

chicken forms of gonadotropin-releasing hormone in the gonads

of a protochordate, Ciona intestinalis Proc Natl Acad Sci USA

97:2343–2348

124 Gorbman A, Sower SA (2003) Evolution of the role of GnRH in

animal (Metazoan) biology Gen Comp Endocrinol 134:207–213

125 Hinsch GW (1980) Effect of mandibular organ implants upon

the spider crab Trans Am Microsc Soc 99:317–322

126 Laufer H, Landau M, Borst D, Homola E (1986) The synthesis

and regulation of methyl farnesoate, a novel juvenile hormone

for crustacean reproduction In: Porchet M, Andries JC,

Dhai-naut A (eds) Advances in invertebrate reproduction, vol 4.

Elsevier, Amsterdam, pp 135–143

127 Rodriguez EM, Lopez Greco LS, Medesain DA, Laufer H,

Fingerman M (2002) Effects of methyl farnesoate alone and in

combination with other hormones on ovarian growth of the red

swamp crayfish, Procambarus clarkii during vitellogenesis Gen

Comp Endocrinol 125:34–40

128 Soroka Y, Sagi A, Khalaila I, Abdu U, Milner Y (2000) Changes

in protein kinase C during vitellogenesis in the crayfish Cherax

quadricarinatus—possible activation by methyl farnesoate Gen

Comp Endocrinol 118:200–208

129 Tiu SH, Chan S, Tobe SS (2010) The effects of farnesoic acid

and 20-hydroxyecdysone on vitellogenin gene expression in the

lobster, Homarus americanus, and possible roles in the

repro-ductive process Gen Comp Endocrinol 166:337–345

130 Chang ES (1997) Chemistry of crustacean hormones that

regu-late growth and reproduction In: Fingerman M, Nagabushanam

R, Thompson MF (eds) Endocrinology and reproduction Recent

advances in marine biotechnology, vol 1 Science Publishers

Inc., New Hampshire, pp 163–178

131 Tamone SL, Chang ES (1993) Methyl farnesoate stimulates

ecdysteroid secretion from crab Y-organs in in vitro Gen Comp

Endocrinol 89:425–432

132 Abdu U, Takac P, Laufer H, Sagi A (1998) Effect of methyl

farnesoate on late larval development and metamorphosis in the

prawn Macrobrachium rosenbergii (Decapoda, Palaemonidae):

a juvenoid effect? Biol Bull 195:112–119

133 Borst DW, Laufer H (1990) Methyl farnesoate, a JH-like

com-pound in crustaceans In: Gupta AP (ed) Recent advances in

comparative arthropod morphology, physiology, and

develop-ment Rudgers University Press, New Brunswick, pp 35–60

134 Marsden G, Hewitt D, Boglio E, Mather P, Richardson N (2008)

Methyl farnesoate inhibition of late stage ovarian development

and fecundity reduction in the black tiger prawn, Penaeus

monodon Aquaculture 280:242–246

135 Homola E, Sagi A, Laufer H (1991) Relationship of claw form

and exoskeleton condition to reproductive system size and

methyl farnesoate in the male spider crab, Libinia emarginata.

Invertebr Reprod Dev 20:219–225

136 Sagi A, Ahl J, Danaee H, Laufer H (1994) Methyl farnesoate

levels in male spider crabs exhibiting active reproductive

behaviour Horm Behav 28:262–272

137 Nagaraju GPC, Reddy PR, Reddy PS (2004) Mandibular organ:

its relation to body weight, sex, molt and reproduction in the

crab, Oziotelphusa senex senex Fabricius 1791 Aquaculture

232:603–612

138 Payen GG, Costlow JD (1977) Effects of a juvenile hormone

mimic on male and female gametogenesis of the mud-crab

Rhithropanopeus harrisii (Gould) (Brachyura: Xanthidae) Biol

Bull 152:199–208

139 Sasikala SL, Subramoniam T (1991) Influence of juvenile mone III (JHIII) on ovarian activity of adult paddy field crab Paratelphusa hydrodromous (Herbst) Indian J Exp Biol 29:426–429

hor-140 Sappington TW, Oishi K, Raikhel AS (2002) Structural acteristics of insect vitellogenin In: Raikhel AS, Sappington

char-TW (eds) Reproductive biology of invertebrates, vol XII (part A) Science Publishers Inc., Enfield, pp 69–101

141 Arvy L, Echalier G, Gabe M (1954) Modifications de la gonade

de Carcinus maenas L apres ablation bilaterale de l’organ Y.

C R Acad Sci Paris Ser D 239:1853–1855

142 Steel CGH, Vafopoulou X (1998) Ecdysteroid titers in lymph and other tissues during molting and reproduction in the terrestrial isopod, Oniscus ascellus (L.) Invertebr Reprod Dev 34:187–194

hemo-143 Charniaux-Cotton H, Touir A (1973) Controlle de nese et de la vitellogenese chez la crevette hermaphrodite Lysmata seticaudata Risso CR Acad Sci Paris D 276:2717–2720

previtelloge-144 Okumura T, Han CH, Suzuki Y, Aida K, Hanyu I (1992) Changes in hemolymph vitellogenin and ecdysteroid levels during the reproductive and non-reproductive molt cycles in the freshwater prawn Macrobrachium nipponense Zool Sci 9:37–45

145 Demeusy N (1962) Role de la gland de mue dans l’evolution ovarianne due crabe Carcinus maenas Linn Can Biol Mar 3:37–56

146 Chaix JC, De Reggi M (1982) Ecdysteroid levels during ovarian development and embryogenesis in the spider crab Acanthonyx lunulatus Gen Comp Endocrinol 47:7–14

147 Young NJ, Webster SG, Rees HH (1993) Ovarian and lymph ecdysteroid titers during vitellogenesis in Macrobrachi-

hemo-um rosenbergii Gen Comp Endocrinol 90:183–191

148 Young NJ, Webster SG, Rees HH (1993) Ecdysteroid profiles and vitellogenesis in Penaeus monodon (Crustacea: Decapoda) Int J Invertebr Reprod 24:107–118

149 Chung ACK, Durica DS, Clifton SW, Roe BA, Hopkins PM (1998) Cloning of crustacean ecdysteroid receptor and retinoid

X receptor gene homologs and elevation of retinoid-X receptor mRNA by retinoic acid Mol Cell Endocrinol 139:209–227

150 Durica DS, Wu X, Anilkumar G, Hopkins PM, Chung ACK (2002) Characterization of crab EcR and RXR homologs and expression during limb regeneration and oocyte maturation Mol Cell Endocrinol 189:56–79

151 Kokoza VA, Martin D, Mienaltowski MJ, Ahmaed A, Morton

CM, Raikhel AS (2001) Transcriptional regulation of the quito vitellogenin gene via a blood meal-triggered cascade Gene 274:47–65

mos-152 Maki A, Sawatsubashi S, Ito S, Shirode Y, Suzuki E, Zhao Y, Yamagata K, Kouzmenko A, Takeyama T, Kato S (2004) Juvenile hormones antagonize ecdysone actions through co-repressor recruitment to EcR/USP heterodimers Biochem Biophys Res Commun 320:262–267

153 Fairs NJ, Evershed RP, Quinlan PT, Goad LJ (1990) Changes in ovarian unconjugated and conjugated steroid titers during vitellogenesis in Penaeus monodon Aquaculture 89:83–99

154 Quinitio ET, Yamauchi K, Hara A, Fuji A (1991) Profiles of progesterone- and estradiol-like substances in the hemolymph of female Penaeus monodon during an annual reproductive cycle Gen Comp Endocrinol 81:343–348

155 Kirubakaran R, Peter SM, Dharani G, Vinithkumar NV, Sreeraj G, Ravindran R (2005) Changes in vertebrate type steroids and 5-hydroxytryptamine during ovarian recrudescence in the Indian spiny lobster Panulirus homarus NZ J Mar Freshw Res 39:527–537

156 Gunamalai V, Kirubagaran R, Subramoniam T (2006) brate steroids and the control of female reproduction in two

decapod crustaceans, Emerita asiatica and Macrobrachium

ro-senbergii Curr Sci 90:119–123

157 Yano I (1985) Induced ovarian maturation and spawning in

greasyback shrimp, Metapenaeus ensis, by progesterone.

Aquaculture 47:223–229

158 Yano I (1987) Effects of 17-hydroxy-progesterone on

vitello-genin secretion in kuruma prawn, Penaeus japonicus

Aqua-culture 61:49–57

159 Yano I, Hoshino R (2006) Effects of 17b-estradiol on the

vitellogenin synthesis and oocyte development in the ovary of

kuruma prawn (Marsupenaeus japonicus) Comp Biochem

Physiol Part A 144:18–23

160 Coccia E, De Lisa E, Di Cristo C, Di Cosmo A, Paolucci M

(2010) Effects of estradiol and progesterone on the reproduction

of the freshwater crayfish Cherax albidus Biol Bull 218:36–47

161 Charniaux-Cotton H, Payen G (1988) Crustacean reproduction.

In: Laufer H, Dower RGH (eds) Endocrinology of selected

invertebrate types Alan R Liss, New York, pp 279–303

162 Fliss AE, Benzeno S, Rao J, Caplan AJ (2000) Control of

estrogen receptor ligand binding by hsp90 J Steroid Biochem

Mol Biol 72:223–230

163 Wu LT, Chu KH (2008) Characterisation of heat shock protein 90

in the shrimp Metapenaeus ensis: evidence for its role in the regulation of vitellogenin synthesis Mol Reprod Dev 75:952–959

164 Paolucci M, Di Cristo C, Di Cosmo A (2002) Immunological evidence for progesterone and estradiol receptors in the fresh- water crayfish Austropotamobius pallipes Mol Reprod Dev 63:55–62

165 Couch EF, Hagino N, Lee JW (1987) Changes in estradiol and progesterone immunoreactivity in tissues of the lobster Homarus americanus with developing and immature ovaries Comp Bio- chem Physiol A87:765–770

166 Preechaphol R, Klinbunga S, Ponza P, Menesveta P (2010) Isolation and characterization of progesterone receptor-related protein p23 (Pm-p23) differentially expressed during ovarian development of the giant tiger shrimp Peneus monodon Aqua- culture 308:S75–S82

O R I G I N A L A R T I C L E Fisheries

Development of Bayesian production models for assessing

the North Pacific swordfish population

Jon Brodziak•Gakushi Ishimura

Received: 29 January 2010 / Accepted: 4 October 2010 / Published online: 12 November 2010

 The Japanese Society of Fisheries Science 2010

Abstract Bayesian surplus production models were

developed for assessing the North Pacific swordfish

popu-lation under alternative scenarios: a single-stock scenario

and a two-stock scenario with subareas that represented the

western central and eastern Pacific Ocean regions Biomass

production was modeled with a three-parameter production

model that allowed production to vary from the symmetric

Schaefer curve using an estimated shape parameter

Log-normal prior distributions for intrinsic growth rate and

car-rying capacity were assumed Goodness-of-fit diagnostics

were developed for comparing the fits of alternative model

configurations based on the root-mean squared error of catch

per unit effort (CPUE) fits and the standardized CPUE

residuals Production model fits for 1952–2006 indicated that

the Japanese longline CPUE numbers were influential under

each stock scenario because these scenarios were the longest

time series of relative abundance indices Model results also

indicated that assumptions about the prior means for intrinsic

growth rate and carrying capacity may be important based on

the model configuration

Keywords Bayesian Swordfish  Production model 

North Pacific

IntroductionSwordfish Xiphias gladius, also known as broadbillswordfish, inhabit a wide region of the Pacific between thelatitudes of 50N and 50S Like other tuna and tuna-likespecies, swordfish is a highly migratory species and isbelieved to have high economic value in both commercialand recreational fisheries [1] In the northern Pacific, theannual total catch has fluctuated around 15,000 t since

2001 The majority of catch has been taken by longlinefishing vessels from Japan, Chinese-Taipei, and the USA(Fig.1), which accounted for 95% of the total harvest inthe North Pacific in 2005, with the remaining catch taken

by Korea and Mexico [International Scientific Committee

of Tuna and Tuna-like Species of the North Pacific Ocean(ISC), Report of the Billfish Working Group Workshop2009: http://isc.ac.affrc.go.jp/pdf/ISC9pdf/Annex_5_ISC9_BILLWG_Feb09.pdf; accessed 24 Sept 2010] Increasinginterest in the harvest of North Pacific swordfish calls for

an appropriate stock assessment, conservation ment, and sustainable development of the fishery How-ever, its high migratory nature and broad geographicdistribution limit the amount of information on spatialpatterns in swordfish biology and stock structure

manage-To date, stock assessments on the North Pacific fish have been conducted assuming a single stock usingcatch, fishery size composition data, and abundance indices[i.e., catch per unit effort (CPUE)] In 2004, Kleiber andYokawa used MULTIFAN-CL to conduct a preliminaryNorth Pacific swordfish in a four-region model (PelagicFisheries Research Program of the Joint Institute forMarine and Atmospheric Research: http://imina.soest.hawaii.edu/PFRP/sctb15/papers/BBRG-3.pdf; accessed 24Sept 2010) In two subsequent studies, Wang et al [2,3]applied a similar age- and length-structured modeling

sword-J Brodziak

NOAA Pacific Islands Fisheries Science Center,

Honolulu, HI 96822-2396, USA

G Ishimura ( &)

Center for Sustainability Science, Hokkaido University,

Sapporo, Hokkaido 060-0809, Japan

e-mail: gakugaku@sgp.hokudai.ac.jp

Fish Sci (2011) 77:23–34

DOI 10.1007/s12562-010-0300-0

approach which included some sex-specific data These

studies concluded that there was little contrast in the North

Pacific swordfish fishery CPUE data to estimate stock

status relative to biological reference points using highly

parameterized age- and length-structured modeling

approaches (ISC Report of the Billfish Working Group

Workshop 2009:http://isc.ac.affrc.go.jp/pdf/ISC9pdf/Annex_

5_ISC9_BILLWG_Feb09.pdf; accessed 24 Sept 2010) In

particular, the relatively flat trends in the swordfish CPUE

and limited fishery length frequency data made it difficult

to estimate the appropriate scale of biomass in the

popu-lation using age-structured models of the popupopu-lation [4]

A revised consideration of the spatial structure of the

swordfish population with updated catch and effort data

and the use of a more parsimonious production modeling

approach with fewer parameters to estimate, however, was

expected to improve model fits to CPUE and to help

esti-mate trends in swordfish abundance and harvest rates

At present, the spatial stock structure of the North

Pacific swordfish population is under investigation The

ISC, which is a scientific body for a regional fishery

management organization that includes the North Pacific,

considered two hypotheses on stock structure according to

available data and studies These two scenarios were: (1)

the single-stock scenario, in which one stock exists in the

North Pacific and (2) the two-stock scenario, in which two

stocks exist, namely, the western central and eastern

trop-ical Pacific stocks, respectively The two-stock scenario

was supported by genetic studies demonstrating significant

spatial heterogeneity in the population genetic structure of

Pacific swordfish [5,6] and by analyses of longline CPUE

showing a boundary in the southeast Pacific [7] As a

result, the two-stock scenario was considered to be the best

working hypothesis for swordfish stock assessment and

management (ISC Report of the Billfish Working Group

Workshop 2009:http://isc.ac.affrc.go.jp/pdf/ISC9pdf/Annex_5_ISC9_BILLWG_Feb09.pdf; accessed 24 Sept 2010).However, both scenarios were analyzed to account for thepossibility that the working hypothesis was incorrectlyaccepted [8]

This study develops a Bayesian statistical framework toestimate parameters of production models to assess theNorth Pacific swordfish population using multinationalfishery catch and effort through 2006 A Bayesian approach

is used so that prior information on probable parametervalues can improve the reliability of estimates of biomass,harvest rate, and biological reference points [9, 10] TheBayesian approach provides direct estimates of parameteruncertainty that are straightforward to interpret for man-agement For example, a 95% Bayesian credibility intervalfor current stock status would include the true value of thestatus parameter with 95% probability of belief In contrast,

a frequentist interpretation of such a confidence intervalwould assign 95% probability to the event that the intervalincludes the true status value over a large number of rep-etitions of the calculations [11] This latter interpretation ofuncertainty in stock status seems more difficult to explain

to managers Bayesian estimates can also be used to struct decision tables where the probabilities of alternativehypotheses and associated performance measures underseparate management actions are shown; such tables canidentify management actions that have acceptably lowprobabilities of undesirable outcomes for precautionarymanagement [12]

con-The production models developed herein include bothprocess error for population biomass dynamics and heter-ogeneous observation errors for fitting the observed CPUEdata from multiple fishing fleets This enables the pro-duction models to incorporate stochastic variation in bio-mass dynamics parameters as well as sampling variation infitting observed CPUE indices Production models wereformulated for two structure-structure scenarios, i.e., thesingle-stock scenario and the two-stock scenario withwestern and eastern stock areas Thus, the overall goal was

to develop operational Bayesian production models thatcould incorporate multiple abundance indices and hetero-geneous observation errors We also discuss the practicalbenefits of applying a Bayesian estimation framework toassess the biological and socioeconomic status of the NorthPacific swordfish resource

MaterialsDataFishery catch data for assessing North Pacific swordfishwere taken from the most recent summary of available

Fig 1 Swordfish landings in the North Pacific by Japan,

Chinese-Taipei, Korea, Mexico, and the USA

fishery-dependent data (D Courtney and L Wagatsuma,

unpublished data, 2009) Commercial catch biomass data

for Japanese, Chinese-Taipei, and Hawaii (USA) longline

fisheries were available for 1951–2006 under each stock

structure scenario (Fig.1), with Japanese vessels

pro-ducing the majority of landings The catch data were

aggregated by region under the two stock-structure

sce-narios supported by the ISC Billfish Working Group The

single-stock scenario assumed that there was one unit

stock of swordfish north of the equator (Fig.2a) The

two-stock scenario assumed that two stocks existed,

separated by a diagonal boundary extending from Baja,

California, to the Equator (Fig.2b), based on the

anal-ysis of longline CPUE by Ichinokawa and Brodziak [7]

along with the analyses of population genetic structure

by Reeb et al [5] and Alvarado Bremer et al [6]

Overall, the catch data were used to model the effects of

fishery removals from the North Pacific swordfish

pop-ulation during 1951–2006

Estimates of standardized commercial fishery CPUEwere provided (D Courtney and L Wagatsuma, unpub-lished data, 2009) for each stock scenario The standard-ized CPUE time series, which consider area and season asmain factors, for the single-stock scenario included Japa-nese longline CPUE (1952–2006, n = 55), Chinese-Taipeilongline CPUE (1995–2006, n = 12), and Hawaii shallow-set longline CPUE (1995–2000 and 2004–2006, n = 9).Changes in the fishery regulations for the shallow-setfishery necessitated that the time series of Hawaii shallow-set data be treated as two separate sets of relative abun-dance indices with set 1 for 1995–2000 and set 2 for2004–2006 Under the two-stock scenario, the set ofavailable standardized CPUE time series was different foreach subarea The standardized CPUE time series for subarea

1 in the western central Pacific included Japanese longlineCPUE (1952–2006, n = 55), Chinese-Taipei longlineCPUE (1995–2006, n = 12), and Hawaii shallow-set long-line CPUE (Set 1, 1995–2000; Set 2, 2004–2006, n = 9)

Fig 2 Assumed stock structure

of the North Pacific swordfish

population: a single-stock

scenario, b two-stock scenario

The standardized CPUE time series for subarea 2 in the

eastern Pacific included Japanese longline CPUE

(1955–2006, n = 52) and Chinese-Taipei longline CPUE

(1995–2006, n = 12)

Methods

Production model and biological reference points

Swordfish production models were formulated as

Bayesian-state space models with explicit observation and process

error terms [13,14] The biomass time series comprised the

unobserved state variables, which were estimated from the

observed relative abundance indices (i.e., CPUE) and

cat-ches using observation error likelihood function and prior

distributions for model parameters (h) In this case, the

observation error likelihood measured the discrepancy

between observed and predicted CPUE, and the prior

dis-tributions represented the relative degree of belief about the

possible values of model parameters

The process dynamics represented the fluctuations in

exploitable swordfish biomass based on density-dependent

processes and fishery harvests The production dynamics of

biomass were based on a power function model with an

annual time step Under this three-parameter model,

cur-rent biomass (BT) depended on the previous biomass

(BT-1), catch (CT-1), intrinsic growth rate (R), carrying

capacity (K), and a production shape parameter (M) for

The production model shape parameter, M, determined

where surplus production peaked as biomass varied as a

fraction of carrying capacity If the shape parameter was

less than unity (0 \ M \ 1), then surplus production

peaked when biomass was below one half of K (i.e., a

left-skewed production curve) If the shape parameter was

greater than unity (M [ 1), then biomass production

peaked when biomass was more than one half of K (i.e.,

a right-skewed production curve) If the shape parameter

was identical to unity (M = 1), then the production model

was identical to a discrete-time Schaefer production model

where maximum surplus production occurred when

biomass was equal to one half of K Thus, the shape of

the biomass production curve could be symmetric or

right-or left-skewed, depending on the estimated value of M

The power function model was reparameterized using

the proportion of carrying capacity (P = B/K) to improve

the efficiency of the Markov Chain Monte Carlo (MCMC)

algorithm used to estimate parameters [13] Given this

parameterization, the process dynamics for the powerfunction model were

Observation error modelThe observation error model related the observed fisheryCPUE to the exploitable biomass of the swordfish stockunder each scenario It was assumed that each CPUE index(I) is proportional to biomass with catchability coefficient Q

The observed CPUE dynamics were subject to naturalsampling variation that was assumed to be lognormallydistributed The observation errors were distributed as

mT ¼ eV T where the VT are independent and identicallydistributed normal random variables with zero mean andweighted variance (WTs)2 with standard deviation s andweighting factor WT The weighting factors (WT) of theannual CPUE variance terms reflected the relativeuncertainty of the value of the CPUE index in year T andwere scaled using the coefficient of variation (CV) of thedifference between the observed and predicted log-transformed biomass indices [15] In particular, theseannual weighting factors were calculated from the relativecoefficients of variation of each annual CPUE index andthe minimum observed CV of CPUE {min[CV(CPUE)]}

as WT= CV(CPUET)/min[CV(CPUE)]

Given the lognormal observation errors, the observationequations for each annual period indexed by T = 1,…, Nwere

IT¼ QKPT mT ð7Þ

This specified the general form of the observation error

likelihood function p(IT|h) for each fishing fleet through

time

Process error model

The process error model related the dynamics of

exploit-able biomass to natural variability in demographic and

environmental processes affecting the swordfish stock The

deterministic process dynamics (Eq.2) were subject to

natural variation as a result of fluctuations in life history

parameters, trophic interactions, environmental conditions,

and other factors In this case, the process error represented

the joint effects of a large number of random multiplicative

events which combined to form a multiplicative lognormal

process under the Central Limit Theorem As a result, the

process error terms were assumed to be independent and

lognormally distributed random variables gT ¼ eU T where

the UT were normal random variables with mean 0 and

variance r2

Given the process errors, the state equations defined the

stochastic process dynamics by relating the unobserved

biomass states to the observed catches and the estimated

population dynamics parameters Assuming multiplicative

lognormal process errors, the state equations for the

ini-tial time period (T = 1) and subsequent periods (T [ 1)

 gT for T [1

ð8ÞThese coupled state equations set the conditional prior

distribution for the proportion of carrying capacity, p(PT),

in each time period T, conditioned on the proportion in the

previous period

Prior distributions

Under the Bayesian paradigm, prior distributions are

employed to quantify existing knowledge, or the lack

thereof, of the likely value of each model parameter For

the production model, the model parameters consisted of

the carrying capacity, the intrinsic growth rate, the shape

parameter, the catchability coefficients, the process and

observation error variances, and the annual biomasses as a

proportion of carrying capacity Auxiliary information was

incorporated into the formulation of the prior distributions

when it was available

Prior for carrying capacityThe prior distribution for the carrying capacity p(K) was alognormal distribution with mean (lK) and variance (rK2)parameters

Prior for intrinsic growth rateThe prior distribution for intrinsic growth rate p(R) was alognormal distribution with mean (lR) and variance (rR2)parameters set to achieve a CV for R of 50%

!

ð10Þ

The mean R parameter was set to be lR= 0.5 for eachstock scenario This mean value was slightly higher thanthe range of prior means (0.40, 0.43) estimated for Northand South Atlantic swordfish, respectively, based on lifehistory parameters [16] A similar analysis used life historyparameters for North Pacific swordfish and the meangeneration time approach, with the results suggesting thathigher mean values of R of approximately 0.9 wereappropriate [17] This analysis assumed the female growthand maturation values of DeMartini et al [18,19] and usedfive alternative natural mortality rate estimators fromBrodziak (ISC Report of the Billfish Working GroupWorkshop 2009:http://isc.ac.affrc.go.jp/pdf/ISC9pdf/Annex_5_ISC9_BILLWG_Feb09.pdf; accessed 24 Sept 2010) tocompute a mean value of R The primary differencebetween the Atlantic and Pacific swordfish life historyparameters was the value of natural mortality McAllister

et al [16] assumed a constant natural mortality rate of

M = 0.2 for Atlantic swordfish, while the Pacific swordfishnatural mortality rate was estimated to be M & 0.35,which is about 75% higher than the Atlantic swordfishvalue While there was uncertainty about an appropriateprior mean for R, setting the prior mean to be lR= 0.5with a CV of 50% allowed sufficient flexibility to estimatethe probable value of R given the observed catch andCPUE data

Prior for production shape parameter

The prior distribution for the production function shape

parameter p(M) was a gamma distribution with scale

parameter k and shape parameter k:

pðMÞ ¼k

kMk1expðkMÞ

The values of the scale and shape parameters were set to

k = k = 2 This choice of parameters set the mean of

p(M) to be lM = 1, which corresponded to the value of M for

the Schaefer production model This choice also implied that

the CV of the shape parameter prior was 71% In effect, the

shape parameter prior was centered on the symmetric

Schaefer model as the default, with sufficient flexibility to

estimate a nonsymmetrical production function if needed

Prior for catchability

The prior for the catchability coefficient p(Q) was chosen

to be a diffuse inverse-gamma distribution with scale

parameter k and shape parameter k

ð12ÞThe scale and shape parameters were set to be

k = k = 0.001 This choice of parameters implied that

1/Q has a mean of 1 and a variance of 1,000, which is a

relatively noninformative prior Since 1/Q is unbounded at

Q = 0, an additional numerical constraint that Q be no

smaller than 0.0001 was imposed for the MCMC sampling

Priors for error variances

Priors for the process error variance p(r2) and observation

error variance p(s2) were chosen to be inverse-gamma

distributions The choice of an inverse gamma distribution

implied that the associated prior for error precision

(p = 1/r2) was effectively p pð Þ / p1, which is the

Jef-frey’s prior for the precision parameter [20] As a result,

inferences based on the gamma assumption were scale

invariant and would not be affected by changing the scale

of the variance parameter For the process error variance

prior, the scale parameter was set to k = 4 and the shape

parameter was k = 0.1 This choice of parameters

pro-duced an expected value of approximately E[r2] = 0.025

with a CV of 16% Similarly, for the observation error

variance prior, the scale parameter was set to k = 2, and

the shape parameter was k = 0.1 This choice of

parame-ters produced an expected value of approximately

E[s2] = 0.05 with a CV of 22% (recall that the annual

observation errors also included a year-specific weighting

factor) Given these prior assumptions, the initial vation error variance was 50% greater than the processerror variance Of course, the estimated values of processand observation error from the MCMC sampling depended

obser-on the model fits to the observed data

Priors for proportions of carrying capacityPrior distributions for the time series of the proportion ofbiomass to carrying capacity, p(PT), were lognormal dis-tributions, as specified in the process dynamics The meanproportion of carrying capacity for the initial year of 1951was set to 0.9 under each stock structure scenario Thiswould correspond to an assumption that the North Pacificswordfish population was lightly exploited and had bio-mass near its carrying capacity following the near cessation

of directed fishing during World War II Sensitivity yses on the influence of the initial proportion of carryingcapacity indicated that altering the assumed value of p(PT)from 0.6 to 1.0 had a minor effect on the scale of estimates

anal-of exploitable biomass

Posterior distributionThe joint posterior distribution of the swordfish productionmodel needs to be sampled in order to make inferencesabout the estimates of the model parameters Given thecatch and the CPUE data, D, the posterior distributionp(h|D) was proportional to the product of the prior distri-butions and the likelihood of the CPUE data via Bayes’theorem

pðhjDÞ / p Kð Þp Rð Þp Mð Þp Qð Þp r 2

 p s  Y2 N

T¼1

p Pð TÞYN T¼1

p IðTjhÞ ð13Þ

Parameter estimation for this nonlinear multiparametermodel was based on generating a large number ofindependent samples from the posterior distribution In thiscase, MCMC simulation using Gibbs sampling was applied

to numerically generate a sequence of samples from theposterior distribution [21] The WINBUGS software(The Bugs Project: http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/manual14.pdf; accessed 24 Sept 2010’’) was used

to set the initial conditions, perform the MCMC calculations,and summarize the results [22]

MCMC simulations were conducted in an identicalmanner for each of the swordfish stock structure scenariosmodels described below Three chains of 60,000 sampleswere simulated for each model A burn-in period of 10,000samples was removed from each chain to remove anydependence of the MCMC samples on the initial condi-tions Next, each chain was thinned by 2 to reduce

autocorrelation, and every other sample was used for

inference As a result, 75,000 samples from the posterior

were used for summarizing model results Convergence of

the MCMC simulations to the posterior distribution was

checked using the potential scale reduction factor [23] and

the Heidelberger and Welch stationarity test [24] These

convergence diagnostics were monitored for several key

model parameters (intrinsic growth rate, carrying capacity,

production function shape parameter, catchability

coeffi-cients, and error variances) to verify convergence of the

MCMC chains to the posterior distribution

Goodness-of-fit criteria

Model residuals were used to measure the goodness-of-fit

of the alternative production models Residuals for the

CPUE series are the log-scale observation errors eT

eT ¼ ln Ið Þ  ln QKPT ð TÞ ð14Þ

Nonrandom patterns in the residuals indicated that the

observed CPUE did not conform to one or more model

assumptions The root mean-squared error (RMSE) of the

CPUE fit provided another diagnostic of the model

goodness-of-fit with lower RMSE indicating a better fit

when models with the same number of parameters were

compared

Results

Convergence to posterior distribution

The potential scale reduction factor was calculated for the

intrinsic growth rate, carrying capacity, production

func-tion shape parameter, catchability coefficients, and error

variance parameters under each stock structure scenario

For all parameters, the estimated potential scale reduction

values were approximately unity which was consistent with

the convergence in distribution of the MCMC samples to

the posterior distribution Similarly, the Heidelberger and

Welch stationarity test could not reject the hypothesis that

the MCMC chains were stationary at the 95% confidence

level for any of the parameters In addition, empirical

examination of the Monte Carlo error, a measure of the

variability in each estimate due to simulation, indicatedthat these errors were relatively small—on the order of0.4–2.5%—of the estimated standard deviation for allparameters This was also consistent with convergence ofthe MCMC chains to the posterior distributions Last,visual inspection of density plots of the posterior distri-butions of the intrinsic growth rate, carrying capacity,production function shape parameter, catchability coeffi-cients, and error variances indicated that these densitieswere smooth and unimodal for all parameters, as expectedfor a convergent sequence of MCMC samples Overall, theconvergence diagnostics that were examined indicated thatthe MCMC samples generated from the production modelhad numerically converged to the posterior distribution.Single-stock scenario model fits to CPUE

Results of the fits to standardized CPUE under the stock scenario indicated that the Japanese longline CPUEhad the lowest RMSE while the Chinese-Taipei longlineCPUE had the poorest fit (Table1) Predicted Japa-nese CPUE appeared to randomly fluctuate about theobserved CPUE time series (Fig.3a) Examination of theJapanese log-scale residuals indicated that there was amoderate but nonsignificant increasing trend with time(P = 0.14) The residuals were normally distributed(P = 0.77) and had constant variance (P = 0.53) The fit tothe observed Chinese-Taipei longline CPUE had a pattern ofconsecutive negative residuals that did not appear to berandom (Fig.3b) There was a nonsignificant increasingtrend in residuals (P = 0.06) with constant variance(P = 0.96) However, the log-scale residuals were not nor-mally distributed (P = 0.04) Similarly, the fits to the Hawaiishallow-set longline CPUE had a negative followed by apositive run of residuals (Fig.3c), but no trends in residualswere detected during 1995–2000 (P = 0.14) or 2004–2006(P = 0.13) The log-scale residuals were normally distrib-uted during 1995–2000 (P = 0.41) with constant variance(P = 0.06), but they were not normally distributed(P \ 0.01) and did have constant variance (P \ 0.01) during2004–2006 Overall, under the single-stock scenario, the fits

single-to the CPUE time series appeared single-to be adequate, althoughthere was some lack of conformance to model errorassumptions in a few cases

Table 1 Root mean-squared errors of model fits to CPUE time series under the single-stock and two-stock scenarios

longline

Chinese-Taipei longline

Hawaii longline shallow—set 1

Hawaii longline shallow—set 2

CPUE Catch per unit effort

Two-stock scenario model fits to CPUE

For the subarea 1 under the two-stock scenario, results of the

fits to standardized CPUE indicated that the Japanese

longline CPUE showed the lowest RMSE, while theChinese-Taipei longline CPUE showed the highest RMSE(Table1) Predicted Japanese CPUE fluctuated around theobserved CPUE time series (Fig.4a) The log-scale residualshad no time trend (P = 0.53), were normally distributed(P = 0.43), and had constant variance (P = 0.75) TheChinese-Taipei longline CPUE fit had a pattern of consec-utive negative residuals in the late 1990s (Fig.4b) Therewas a detectable time trend in the residuals (P = 0.02), andthe log-scale residuals were normally distributed (P = 0.33)and had constant variance (P = 0.12) Fits to the Hawaiishallow-set longline CPUE appeared to have an increasingtrend in residuals (Fig.4c) There was a significantincreasing trend during 1995–2000 (P = 0.02) and a non-significant increasing trend during 2004–2006 (P = 0.07).The log-scale residuals were normally distributed during1995–2000 (P = 0.18) but were not normally distributedduring 2004–2006 (P \ 0.01); the log-scale residuals hadconstant variance during 1995–2000 (P = 0.06) but notduring 2004–2006 (P \ 0.01) Overall, some of the fits to theCPUE time series in subarea 1 did not appear to be randomand, in particular, the Chinese-Taipei and Hawaii shallow-set long CPUE fits showed increasing trends in their residualpatterns

For subarea 2 under the two-stock scenario, the modelfits to standardized CPUE indicated that the Japaneselongline CPUE had a lower RMSE than did the fit to theChinese-Taipei CPUE (Table1) The fit to the Japaneselongline CPUE (Fig.5a) showed some large negativeresiduals in the 1950s but otherwise appeared to fluctuaterandomly about the observed CPUE The residuals had amoderate significant time trend (P = 0.01), but the log-scale residuals were not normally distributed (P \ 0.01)and the variance was not constant (P \ 0.01) In contrast,there was no apparent pattern in the fit to the Chinese-Taipei longline CPUE (Fig.5b) In this case, the residualshad no detectable trend (P = 0.59), the log-scale residualswere normally distributed (P = 0.68), but the variance wasnot constant (P \ 0.01) Overall, in subarea 2, there was agood fit to the Chinese-Taipei longline CPUE and somelack of fit to the Japanese longline CPUE in the 1950s.Estimates of model parameters and reference pointsEstimates of production model parameters varied betweenthe stock structure scenarios (Table 2) Under the single-stock scenario, the intrinsic growth rate was estimated to be

R = 0.61 In contrast, under the two-stock scenario, theestimates of R were 0.55 and 0.41 for subareas 1 and 2,respectively, or 10 and 33%, respectively, below the single-stock estimate The estimate of K under the single-stockscenario (K = 127.6 kt) was about 31% less than the sum

of the estimates of K under the two-stock scenario

Observed Shallow Set 1

Observed Shallow Set 2

Predicted Shallow Set 1

Predicted Shallow Set 2

(a)

(b)

(c)

Fig 3 Time series of observed and predicted catch per unit effort

(CPUE) of swordfish under the single-stock scenario: a Japanese

longline during 1952–2006, b Chinese-Taipei longline during

1995–2006, c Hawaii shallow-set longline during 1995–2000 and

2004–2006

(K1? K2= 185.0 kt) The estimate of the production

model shape parameter for the single-stock scenario was

M = 1.24, indicating a left-skewed production curve

In comparison, the estimate of M1 for subarea 1 was

approximately 0.93, indicating an approximately symmetric

biomass production curve, while the M2estimate for subarea

2 was M2= 0.59, indicating a right-skewed productioncurve Overall, estimates of production model parameters R,

K, and M differed among the stock scenarios

Estimates of biological reference points also differedamong the stock scenarios (Table2) The mean estimate of

BMSYunder the single-stock scenario was BMSY= 65.1 kt.This was about 26% below the sum of the estimates of

BMSY under the two-stock scenario The mean estimate of

HMSYunder the single-stock scenario was HMSY= 0.29 Incomparison, the estimates of HMSY under the two-stockscenario were 0.23 and 0.13 for subareas 1 and 2, respec-tively, or respectively 21 and 55% less than the single-stock estimate In contrast, the mean estimate of MSYunder the single-stock scenario was MSY = 18.2 kt, whichwas only 6% higher than the sum of the MSY estimatesunder the two-stock scenario

In contrast to the estimates of production modelparameters and biological reference points, there was nopractical difference in the estimates of stock status in 2006

(a)

(b)

(c)

Fig 4 Time series of observed and predicted CPUE of swordfish in

subarea 1 under the two-stock scenario: a Japanese longline during

1952–2006, b Chinese-Taipei longline during 1995–2006, c Hawaii

shallow-set longline during 1995–2000 and 2004–2006

Year

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

0.0 0.1 0.2 0.3 0.4 0.5

0.6

Observed Predicted

Year

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

0.0 0.1 0.2 0.3 0.4 0.5

0.6

Observed Predicted

(a)

(b)

Fig 5 Time series of observed and predicted Japanese CPUE of swordfish in subarea 2 under the two-stock scenario: a Japanese longline during 1955–2006, b Chinese-Taipei during 1995–2006

between the two-stock scenarios (Table2) In particular,

the mean estimates of B2006were greater than BMSYunder

both stock scenarios and subareas, and the associated

probabilities of B2006 exceeding BMSY were 1, except for

subarea 1 where that probability was 0.89 Similarly, mean

estimates of exploitation rate in 2006 were below HMSYfor

both stock scenarios and subareas, and the corresponding

probabilities that H2006exceeded HMSYwere [0.01

Estimates of exploitable biomass and exploitation rate

Under the single-stock scenario, exploitable biomass

fluc-tuated about BMSY during the 1950s to 1980s (Fig.6a)

Biomass then increased above BMSY in the late 1980s,

subsequently declining to below BMSYin the late 1990s and

increasing to above BMSYin the 2000s Exploitation rates

were below HMSYin the early 1950s, increasing to a peak

of about 45% around 1960 and subsequently declining to

less than HMSY during 1965–1990 (Fig.6b) Exploitation

rates increased in the early 1990s to fluctuate around HMSY,

subsequently declining in the early 2000s to roughly half of

HMSY Under the single-stock scenario, exploitable

bio-mass generally remained above BMSY, and exploitation

rates remained below HMSY throughout the assessment

time horizon

Exploitable biomass of the swordfish stock in subarea

1 under the two-stock scenario also fluctuated around

BMSY for most of the assessment time horizon (Fig.6c)

Biomass increased to above BMSY during 1985–1995 and

has since declined to roughly BMSY Exploitation rates in

subarea 1 increased from low values in the 1950s to a

peak of about 40% around 1960 and then declined to

fluctuate about half of HMSY from the mid 1960s to the

late 1980s Exploitation rates increased to fluctuate about

HMSY during the 1990s and then declined in the 2000s to

about two thirds of HMSY Overall, exploitable biomass

in subarea 1 remained at or above BMSY, while

exploi-tation rates remained at or below HMSY throughout the

assessment time horizon

Exploitable biomass in subarea 2 under the two-stock

scenario was at or above BMSYthroughout the assessment

time horizon (Fig.6e) Biomass increased to a peak around

2000 and has since declined in the 2000s, albeit to a levelwell above BMSY Exploitation rates in subarea 2 remained

at or below HMSYthroughout the assessment time horizon(Fig.6f) Overall, the stock in subarea 2 has not beendepleted or experienced overfishing under this modelscenario

DiscussionThe aim of this study was to develop a Bayesian statisticalframework to estimate parameters of production models toassess the North Pacific swordfish under two alternativestock-structure scenarios Overall, the results suggest thatthe North Pacific swordfish population would be estimated

to be a smaller (lower carrying capacity K) and moreproductive stock (higher intrinsic growth rate R) under thesingle-stock scenario than as a combination of two stocksunder the two-stock scenario Similar results were foundfor the estimated biological reference points under the twoscenarios, which indicated that the choice of stock scenariohad no practical impact on the status of the North Pacificswordfish population with respect to MSY-based referencepoints The MSY results from the two stock scenariossuggest that the North Pacific swordfish population is fairlyresilient to fishing pressures and that current biomass isclose to the level of BMSY As a result, the North Pacificswordfish population was not depleted with a high degree

of confidence Under both stock structure scenarios,swordfish exploitation rates were estimated to haveremained below HMSY throughout the assessment timehorizon This implies that the current fishing effort is likelysufficient to conserve the North Pacific swordfish stock(s)while providing for a sustainable fishery Subsequent to ouranalyses, stock synthesis models were constructed to assessthe North Pacific swordfish population, and the stock statusresults were similar to those reported here (ISC Report ofthe Billfish Working Group Workshop 2009: http://isc.ac.affrc.go.jp/pdf/ISC9pdf/Annex_5_ISC9_BILLWG_Feb09.pdf; accessed 24 Sept 2010)

In our study, the Bayesian estimation approach provided

a consistent theory for providing scientific advice that

Table 2 Mean estimates of production model parameters under the single-stock and two-stock scenarios

Stock scenario Mean

R

Mean K

Mean M

Mean

B2006

Pr (B2006[ B MSY )

Mean

H2006

Pr (H2006[ H MSY ) Single-stock scenario 0.61 127.6 1.24 65.1 0.29 18.2 103.8 1.00 0.12 0.00

Two-stock scenario subarea 1 0.55 128.0 0.93 62.2 0.23 14.0 78.0 0.89 0.14 0.01

Two-stock scenario subarea 2 0.41 57.0 0.59 25.6 0.13 3.2 52.8 1.00 0.04 0.01

R, Intrinsic growth rate; K, carrying capacity; M, production model shape parameter; BMSY, biomass to produce maximum sustainable yield;

HMSY, exploitation rate to produce maximum sustainable yield; MSY, maximum sustainable yield; B2006, exploitable biomass in 2006;

Pr (B2006[ B MSY ), probability that B2006exceeds BMSY; H2006, exploitation rate in 2006; Pr (H2006[ H MSY ), probability that H2006exceeds

HMSY

accounted for uncertainty in estimates of stock status

rel-ative to biological reference points These benefits are

considered to be important for effectively conveying stock

assessment results to fisheries managers and stakeholders

Using a Bayesian estimation approach allowed us to make

clear statements about the degree of confidence anduncertainty in estimated quantities [11] In practice, theprobabilistic interpretation of stock status showed that itwas very likely that the swordfish population biomass wasabove BMSY in 2006, with a frequency of [9 out of 10 in

Harvest Rate 95% CI

Year

1950195519601965197019751980198519901995200020052010

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

H MSY

Harvest Rate 95% CI

Biomass 95% CI

Harvest Rate 95% CI

Fig 6 Trends in exploitable biomass (1000 t) and exploitation rate

of North Pacific swordfish under the single-stock scenario,

1951–2006: a exploitable biomass (1000 t) under the single-stock

scenario, b exploitation rate in subarea under the single-scenario,

c exploitable biomass (1000 t) in subarea 1 under the two-stock

scenario, d exploitation rate in subarea 1 under the two-stock scenario, e exploitable biomass (1000 t) in subarea 2 under the two- stock scenario, f exploitation rate in subarea 2 under the two-stock scenario 95% CI 95% Confidence interval

similar circumstances Similarly, it was extremely unlikely,

with a frequency of\1 out of 100 in similar circumstances,

that the swordfish population was being fished in excess of

HMSYin 2006, regardless of the stock-structure scenario In

addition, the use of a Bayesian approach would be

expected to help with the implementation of a

precau-tionary approach to swordfish fishery management in

which managers could choose acceptable risk levels for

undesirable outcomes for the swordfish fishery system

using decision tables to judge the efficacy of alternative

management options [12, 16] Despite these benefits, it is

important to note that the choice of prior distributions can

alter posterior estimates of stock status, especially when

data quality is questionable [25] In this context, it is very

useful to select priors that are consistent with data from

other populations, as was done in this analysis for the prior

distribution of swordfish intrinsic growth rate

We conclude from our development of a Bayesian

production model for the North Pacific swordfish that it is

possible to produce parameter estimates and credibility

intervals to measure changes in exploitable biomass and

harvest rate through time Although the results suggest that

the North Pacific swordfish population is not currently

depleted, the rising economic incentives for increased

harvesting of swordfish may increase the level of fishing

effort on the population, possibly leading to unsustainable

fisheries In this context, our probabilistic results could be

used to provide estimates of the potential fishing capacity

for North Pacific swordfish along with an appropriate

characterization of the uncertainty in such estimates [26]

We also recommend that further assessment work on North

Pacific swordfish should consider more detailed biological

data with age- or length-structured models and also provide

the capacity to make stochastic catch projections under

alternative harvest scenarios

References

1 Ward P, Porter JM, Elscot S (2000) Broadbill swordfish: status of

established fisheries and lessons for developing fisheries Fish

Fish 1:317–336

2 Wang SP, Sun CL, Punt A, Yeh SZ (2005) Evaluation of a

sex-specific age-structured assessment method for the swordfish,

Xiphias gladius, in the North Pacific Ocean Fish Res 73:79–97

3 Wang SP, Sun CL, Punt A, Yeh SZ (2007) Application of the

sex-specific age- structured assessment method for swordfish,

Xiphias gladius, in the North Pacific Ocean Fish Res 84:282–300

4 Ludwig D, Walters C (1985) Are age-structured models

appro-priate for catch-effort data? Can J Fish Aquat Sci 42:1066–1072

5 Reeb C, Arcangeli L, Block B (2000) Structure and migration

corridors in Pacific populations of the swordfish Xiphias gladius,

as inferred through analyses of mitochondrial DNA Mar Biol

136:1123–1131

6 Alvarado BJ, Hinton M, Greig T (2006) Evidence of spatial genetic heterogeneity in Pacific swordfish (Xiphias gladius) revealed by the analysis of LDH-A sequences Bull Mar Sci 79:493–503

7 Ichinokawa M, Brodziak J (2010) Using adaptive area cation to standardize catch rates with application to North Pacific swordfish (Xiphias gladius) Fish Res doi: 10.1016/j.fishres 2010-08-001

stratifi-8 Waples R (1998) Separating the wheat from the chaff: patterns of genetic differentiation in high gene flow species J Hered 89:438–450

9 Hoenig J, Warren W (1994) Bayesian and related approaches to fitting surplus production models Can J Fish Aquat Sci 51:1823–1831

10 McAllister M, Kirkwood G (1998) Bayesian stock assessment: a review and example application using the logistic model ICES J Mar Sci 55:1031–1060

11 Ellison A (2004) Bayesian inference in ecology Ecol Lett 7:509–520

12 Hilborn R, Peterman R (1996) The development of scientific advice with incomplete information in the context of the pre- cautionary approach FAO Fish Tech Pap No 350/2 Food and Agriculture Organization of the UN, Rome, pp 77–102

13 Meyer R, Millar R (1999) BUGS in Bayesian stock assessments Can J Fish Aquat Sci 56:1078–1086

14 Brodziak J (2007) An investigation of alternative production models to assess the Hawaiian bottomfish complex Pacific Islands Fish Sci Cent Admin Rep H-07-01 Pacific Islands Fish- eries Science Center, National Marine Fisheries Services NOAA, Honolulu

15 Maunder M, Starr P (2003) Fitting fisheries models to ized CPUE abundance indices Fish Res 63:43–50

standard-16 McAllister M, Babcock E, Pikitch E, Prager M (2000) tion of a non-equilibrium generalized production model to South and North Atlantic swordfish: combining Bayesian and demo- graphic methods for parameter estimation Col Vol Sci Pap ICCAT 51:1523–1550

Applica-17 McAllister M, Pikitch E, Babcock E (2001) Using demographic methods to construct Bayesian priors for the intrinsic rate of increase in the Schaefer model and implications for stock rebuilding Can J Fish Aquat Sci 58:1871–1890

18 DeMartini E, Uchiyama J, Williams H (2000) Sexual maturity, sex ratio, and size composition of swordfish, Xiphias gladius, caught by the Hawaii-based pelagic longline fishery Fish Bull 98:489–506

19 DeMartini E, Uchiyama J, Humphreys R, Sampaga J, Williams H (2007) Age and growth of swordfish (Xiphias gladius) caught by the Hawaii-based pelagic longline fishery Fish Bull 105:356–367

20 Congdon P (2001) Bayesian statistical modeling Wiley, New York

21 Gilks WR, Richardson S, Spiegelhalter DJ (1996) Markov Chain Monte Carlo in practice Chapman and Hall, London

22 Lunn DJ, Thomas A, Best N, Spiegelhalter D (2000) BUGS—a Bayesian modelling framework: concepts, structure, and extensibility Stat Comput 10:325–337

Win-23 Gelman A, Rubin D (1992) Inference from iterative simulation using multiple sequences Stat Sci 7:457–511

24 Heidelberger P, Welch P (1992) Simulation run length control in the presence of an initial transient Oper Res 31:1109–1144

25 Booth A, Quinn TII (2006) Maximum likelihood and Bayesian approaches to stock assessment when data are questionable Fish Res 80:169–181

26 Arrizabalaga H, Restrepo V, Maunder M, Majkowski J (2009) Using stock assessment information to assess fishing capacity of tuna fisheries ICES J Mar Sci 66:1959–1966

O R I G I N A L A R T I C L E Fisheries

Productive efficiency of the sandfish Arctoscopus japonicus

coastal gillnet fishery using stochastic frontier analysis

Do-Hoon Kim•Kyoung-Hoon Lee•

Bong-Seong Bae•Seong-Wook Park

Received: 18 May 2010 / Accepted: 18 October 2010 / Published online: 23 November 2010

 The Japanese Society of Fisheries Science 2010

Abstract It is important to estimate the productive

effi-ciencies of industries, especially the fishing industry, in

order to determine policies that can improve business

conditions In this study, the productive efficiency of the

sandfish coastal gillnet fishery on the east coast of Korea

has been estimated using stochastic frontier analysis (SFA)

A translog production function wherein the inefficiency

was represented by a truncated-normal distribution was

established; the output variable was the trip production

quantity, the input variables were physical production

factors directly related to the fishing activities of vessels,

such as tonnage, horsepower, and the number of employed

fishers The average productive efficiency of the sample

was 0.59 [0.40–0.79], which implied that productive

inef-ficiency occurs in sandfish coastal gillnet vessels

More-over, it was verified that there are no differences among

the average productive efficiencies of fishing vessels of

different tonnages

Keywords Productive efficiency Stochastic frontieranalysis  Coastal gillnet fishery  Sandfish  Productivity

IntroductionProductive efficiency (PE) is a measure of the ability ofproducers to maximize their output using a given set ofinputs and technologies [1], and it is an important index forassessing and estimating the productivity or profitability of

a producer Producers can compare measured productiveefficiencies, and they can use the results as data for diag-nosing and improving their business conditions Moreover,these results can also be used by an entire industry—forexample, the fishing industry—as useful data for formu-lating policies to foster its development and to assessglobal competitiveness

Owing to its considerable importance, productive ciency has been estimated for many industries, such as thebanking sector, hospital businesses, manufacturing indus-tries, and the tourism sector [2 6] In the fishing industry,the productive efficiencies of the Hawaii longline fishery[7], the scallop fishery in the central Atlantic Ocean [8],and the purse seine fishery in the Gulf of Ca´diz [9] havebeen estimated In particular, in recent years, the impor-tance of estimating the productive efficiency of the fishingindustry has increased because of deteriorating profits infishing businesses due to depleting fishery resourcescoupled with increasing costs such as the fuel cost.Such estimations can be used to determine methods ofimproving fishing business conditions so that the pro-ductivity and profitability of the fishing industry areenhanced

effi-The estimation of productive efficiency can also be lized effectively to formulate policies for the measurement

uti-D.-H Kim

Technology Management Center, National Fisheries Research

and Development Institute, Busan 619-902, Korea

e-mail: delaware310@nfrdi.go.kr

K.-H Lee ( &)  S.-W Park

Fisheries System Engineering Division, National Fisheries

Research and Development Institute, Busan 619-902, Korea

e-mail: khlee71@nfrdi.go.kr

S.-W Park

e-mail: swp4283@nfrdi.go.kr

B.-S Bae

Aquaculture Industry Division, East Sea Fisheries Research

Institute, Gangneung 210-861, Korea

e-mail: asako@nfrdi.go.kr

Fish Sci (2011) 77:35–40

DOI 10.1007/s12562-010-0308-5

and management of fishing capacity, which is an important

issue both domestically and internationally Further, it can

be applied extensively in impact analyses of new fishery

management policies or technological developments Under

the fish stock rebuilding plan, which is being executed

globally, it is essential to estimate the productive efficiencies

of the related fisheries It is very important to not only pursue

the restoration of fishery resources from a biological

view-point but also measure the productive efficiencies of fishing

businesses, and thus to select economically stable

manage-ment measures based on the results

There are two popular methods of estimating productive

efficiency—data envelopment analysis (DEA), which is a

nonparametric method, and stochastic frontier analysis

(SFA), which is a parametric method using econometric

models Of these methods of measuring the fishing

capacity, DEA is the most widely used, because it can

overcome limitations in the available data, and

measure-ments are relatively easy to perform in this method

[10–13]

Both the DEA and the SFA methods that are used to

analyze production efficiency have strengths and

weak-nesses [14–17] The advantage of the DEA method is that

its analysis can include multiple outputs; however,

inef-ficiency values may be exaggerated in this method as

errors in measurement are included in the inefficiency

Moreover, it is unable to statistically verify the used

variables [18–20] On the other hand, in the SFA method,

not only is it possible to statistically verify the variables

used in the model, but probabilistic error and inefficiency

can also be measured separately However, the form of

the production function and the distribution of the

ineffi-ciency must be assumed in advance [21, 22]

Neverthe-less, these weaknesses can be partially compensated for

by statistical verifications, so they are not really

irresolvable

In this study, the productive efficiency of a sandfish

coastal gillnet fishery on the east coast of Korea was

esti-mated using the SFA method Given the present situation

regarding sandfish resources, a fish stock rebuilding plan

was established in 2006 According to this plan, it is

important to take effective management measures to

enhance the productivity and profitability of sandfish

fish-eries by measuring the production efficiencies of individual

fishing vessels Moreover, the selection and adoption of

resource rebuilding measures are required to stabilize

fishing business conditions This study is expected to

pro-vide policy implications toward that end In this study, an

SFA production function was established based on data

from reports of fishing trips made by sample sandfish

coastal fishing vessels in 2007, and the production

effi-ciency per sample vessel was estimated

Materials and methodsSandfish coastal gillnet fisheryThe sandfish Arctoscopus japonicus is an important com-mercial species that inhabits the east coast of Korea It iscaught by diverse fisheries such as coastal gillnets, easternsea Danish seines, eastern sea trawls, and west southernDanish seines An examination of the trend in sandfishcatches shows that the highest figure (approximately25,000 t) was recorded in 1971 Since then, there has been

a downward trend, and only 2,600 t were caught in theearly 1990s The catches fluctuated in the range ofaround 2,000 t thereafter, and increased to 3,400 t in 2002.Figure1 shows the sandfish catches in the period1993–2008; in 2008, the sandfish catch was 2,700 t

As a result of this sharp decline in the number ofsandfish catches, a fish stock rebuilding plan was formu-lated in 2006 Under this plan, the target was to restoreresources such that a sandfish catch of up to 5,000 t would

be possible by 2015 Further, as management measures torecover sandfish resources, existing restrictions such as sizelimits and closed seasons were applied more strictly, whilemarine protected areas centering on spawning groundswere designated and limitations on the use of fishing gearand the total allowable catch (TAC) were proposed

In terms of sandfish catches by fisheries, the catches ofthe eastern sea Danish seine and coastal gillnet fisheries aredominant In terms of the average fishing catches in

3 years, 2006–2008, the eastern sea Danish seine andcoastal gillnet fisheries caught 1,469 and 1,249 t, respec-tively, accounting for 48 and 41%, respectively, of the totalnumber of sandfish catches However, in terms of thenumber of fishing vessels, in 2008, the permitted number ofeastern sea Danish seines was 42, while that of coastal

-500 1,000 1,500 2,000 2,500 3,000 3,500 4,000

Year

total coastal gillnet

Fig 1 Sandfish catches in the period 1993–2008

gillnets was 4,820; this clearly shows that among the

sandfish catching vessels, the coastal gillnet vessels were

dominant

The coastal gillnet vessels catch various species of fish

such as sandfish, snow crab, atka mackerel, cuttlefish, and

flatfish all the year round However, in coastal gillnet

fish-eries, the main period of sandfish catching is the 3 months

from October to December when sandfish catching is

per-formed very intensively As shown in Fig.1, the sandfish

catch by coastal gillnet fisheries was less than 700 t until

2000, but it rose to 1,129 t in 2002, and thereafter continued

its increasing trend to reach 1,835 t in 2007 However, the

figure decreased again to 770 t in 2008

Analytical data

In this study, the trip reports of 17 sampled fishing vessels

during the main sandfish catching period of October–

December in 2007 were used as data to analyze the

productive efficiency of the sandfish coastal gillnet

fish-ery The trip report contains well-arranged data on the

number of fish catches per trip, dates and hours of net

casting and hauling, and the physical elements of the

fishing vessels

To analyze the productive efficiency, the quantity

pro-duced per trip was selected as the output variable Tonnage,

horsepower, and the number of fishers—i.e., the physical

elements of production directly related to the fishing

activities of coastal gillnet vessels—were chosen as the

input variables, based on previous studies [7 9] The

descriptive statistics of the variables used in the analysis

are listed in Table1

The amount of sandfish caught per trip among the

sampled vessels in the coastal gillnet fishery was 66 kg on

average, and it was also found that the deviations in the

number of catches per trip were relatively large, since they

depended on factors such as the physical characteristics of

vessels and fishing periods Furthermore, analysis of the

other input variables confirmed that the tonnage of a fishing

vessel was 2.6 t, its horsepower was 160 hp, and the

number of fisherman employed per vessel was one, on

average

Stochastic frontier analysis (SFA)

In the SFA method, the relationship between the input andoutput variables is described as the production function,while the error terms comprise a probability error term and

a term that reflects production inefficiency [21, 23, 24]

In other words, the function that yields maximum outputwhen a certain quantity of production elements is input-

ed under the present technology levels is defined as in

Eq 1:

Further, by taking logs on both sides of the productionfunction in Eq.1, the SFA production function can beobtained, as shown in Eq.2:

y0i¼ f ðx0i;bÞ þ vi ui: ð2ÞHere, y0i¼ log yi and x0i¼ log xi; yi denotes theproduction of the ith producer (i = 1, 2,…,n); xi is a(1 9 k) vector of the input element function of the ithproducer; and b is a (k 9 1) vector of the parameters to beestimated Furthermore, vi is assumed to be anindependently and identically distributed random error,while uiis a non-negative random variable that is associatedwith the technical inefficiencies of the production of the ithproducer

In the case of the distribution of the inefficiency variable,

4 types of distribution—half-normal, truncated-normal,exponential, and gamma distributions—have been pro-posed, and to estimate parameters, the maximum likelihoodmethod can be applied [21,23–25]

Assuming that the inefficiency variable, ui, follows thetruncated-normal distribution ½Nþðl; r2Þ, the case of

ui= 0 implies that efficient production is implemented,while ui[ 0 implies that inefficient production is imple-mented [25] If the truncated-normal distribution isassumed, the expected value ½EðuijeiÞ of the conditionaldistribution of inefficiency uito the composed error terms(ei= vi- ui) is induced, as shown in Eq.3 [21,23–25]:EðuijeiÞ ¼ rk

Eq 4:

The maximum (frontier) production for the ith producer

is computed as its actual production divided by itsproduction efficiency estimate

Table 1 Summary statistics for the variables used in productive

Results of productive efficiency estimation by SFA

For the estimation of the productive efficiency of sandfish

coastal gillnet vessels, the translog production function was

assumed to take the form of the production function, as

shown in the following equation:

ln catchi¼ b0þ b1ln toniþ b2ln hpiþ b3ln empi

þ b4ðln tonÞ2i þ b5ðln hpÞ2i þ b6ðln empÞ2i

þ b7ðln tonÞðln hpÞiþ b8ðln tonÞðln empÞi

þ b9ðln hpÞðln empÞiþ vi ui: ð5Þ

In the function, i is the ith vessel of the sample vessels;

the output variable is catch, a production quantity; and the

input variables are ton, hp, and emp, which are the number

of tons, horsepower, and the number of people employed,

respectively Further, v and u are the independently and

identically distributed random error and inefficiency,

respectively, as mentioned previously

The results of production function estimation by the SFA

method, where it was assumed that the inefficiency variable

uifollows a truncated-normal distribution [Nþðl; r2Þ], are

shown in Table2

From the analysis, the coefficients for all of the variables

were estimated to be statistically significant within the 1%

level, except for one variable, (ln emp)2 Further, the sum

of the error variances (r2) was found to be statistically

significant, while among the error terms, ruof the

ineffi-ciency variable was not statistically zero; it was verified to

be significant after making the assumption that the

ineffi-ciency evaluation was good

In particular, the weight (k) for the part described by theproductive inefficiency was found to be 1.91, which isstatistically significant Therefore, the part of the produc-tive inefficiency in the error term of the assumed produc-tion function is important for determining the degree andrange of production of a vessel used for sandfish coastalgillnet fisheries

The null hypothesis that the half-normal distribution is

an adequate representation of the distribution of the ficiency effects (H0: l = 0) was rejected; this implies thatthe truncated-normal distribution is an adequate represen-tation of the distribution of the inefficiency effects Thegeneralized likelihood ratio test of the one-sided error term(H0: ru= 0) indicated that technical inefficiency wassignificant In addition, the generalized likelihood ratio testindicates that the translog functional form rather than theCobb–Douglas production functions should be selected forthe SFA of production, as shown in Table3 [26–28].The average productive efficiency of sandfish coastalgillnet vessels was estimated by SFA to be 0.59, and thedistribution of the estimated values of the productive effi-ciency (PE) for the sample vessels is shown in Fig.2 Morespecifically, Table 4 shows the frequency distribution ofthe productive efficiencies for sample vessels The differ-ence in the productive efficiencies per vessel was found out

inef-to be significant, based on the result that the number of

Table 2 Parameter estimates from stochastic frontier analysis (SFA)

of the productive efficiency (PE)

Variable Parameter Coefficient Std.

error

Asymptotic T-ratio

Vessel Fig 2 Productive efficiency (PE) distribution for sample vessels

Table 3 Generalized likelihood ratio tests of hypotheses for eters of the SFA production function

param-Null hypothesis Likelihood

ratio

Critical value (5%)

Critical value (1%)

vessels with PE = 1 was none, the number of vessels with

PE [ 0.6 was 7 (41.2%), while the number of vessels with

PE \ 0.6 was 10 (58.8%) For a more detailed review of

the productive efficiency per sample vessel of sandfish

coastal gillnet fisheries, we can consider the differences in

the productive efficiency among trips per vessel However,

the average productive efficiency was estimated to be in the

range of 0.79–0.40

ANOVA analysis was subsequently conducted in order

to verify the differences in the productive efficiency in

terms of fishing vessel tonnage on the basis of the

esti-mated results of the productive efficiency per sample

ves-sel In the analysis, the vessel tonnage was divided into

three groups: \1, 1–3, and [3 t The results of the analysis

showed that the average productive efficiencies were 0.56

and 0.60 for vessels with tonnages of \1, 1–3 t, and 0.59

for those [3 t Further, the average productive efficiency

per vessel tonnage was not significant statistically, which

confirmed that the productive efficiency by vessel tonnage

did not vary, as shown in Table5

Discussion

In this study, the productive efficiency of coastal gillnet

fishing vessels, which is the primary fishery involved in

catching sandfish, one of the target species in the Korean

fish stock rebuilding plan, was estimated by the SFA

method In the SFA method, the translog production

function—wherein the inefficiency term is assumed to have

a truncated-normal distribution—was used Furthermore,the output variable was the quantity produced per trip, andthe input variables were physical production factors: ton-nage, horsepower, and the number of fishers employed.From the productive efficiency estimation, the averageefficiency of the sample vessels in the sandfish coastalgillnet fishery was found to be 0.59 [range 0.40–0.79],leading to the conclusion that productive inefficiencyoccurs in the vessels The productivity and profitability ofcoastal gillnet fisheries would significantly increase if thevessels were to achieve efficient production Further, it wasverified that the average productive efficiency by tonnage

of fishing vessel does not vary

In addition, the SFA results provide useful policyimplications for sandfish resource management In otherwords, to successfully achieve the sandfish stock rebuildingplan, measures for reducing and managing the fishingcapacities of coastal gillnet vessels should be considered;this should be done in addition to previously recommendedrecovery measures, such as establishing marine protectedareas and limiting the use of fishing gear and TAC Forinstance, vessel buyback programs and limitations on thehorsepower and tonnage of vessels can be proposed toreduce fishing capacity levels and improve the fishingbusiness conditions for coastal gillnet vessels

In this study, the productive efficiency of sandfishcoastal gillnet vessels was analyzed using data coveringonly 1 year due to limitations in the available data; hence,dynamic changes in the productive efficiency over timecould not be estimated The results of this study are notsufficient to generalize the productive efficiency levels ofsandfish coastal gillnet fishing vessels An analysis of theeffect of the financial characteristics of the vessels and thefeatures of their captains will also help to enhance theirproductive efficiency If, in the future, dynamic analyses ofproductive efficiency changes are conducted or its rela-tionships with diverse characteristics are researched, it will

be possible to formulate more useful policy implicationsthat should lead to the enhanced productivity and profit-ability of individual fishing vessels under the sandfish stockrebuilding plan

Acknowledgments We would like to thank the captains of the fishing vessels for their assistance in providing necessary information This study was supported in part by a grant (RP-2010-EC-005) pro- moted by the National Fisheries Research & Development Institute of the Republic of Korea.

References

1 Fried H, Lovell C, Schmidt S (1993) The measurement of ductive efficiency: techniques and applications Oxford Univer- sity Press, New York

pro-Table 4 Frequency distribution of productive efficiency from SFA

Efficiency score Number of vessels

Standard deviation

F statistics (P value)

2 Hofmarcher M, Paterson I, Riedel M (2002) Measuring hospital

efficiency in Austria: a DEA approach Health Care Manag Sci

5:7–14

3 Hwang S, Chang T (2003) Using data envelopment analysis to

measure hotel managerial efficiency change in Taiwan Tour

Magt 24:357–369

4 Madlener R, Antunes C, Dias L (2009) Assessing the

perfor-mance of biogas plants with multi-criteria and data envelopment

analysis Eur J Oper Res 197:1084–1094

5 Olesen O, Petersen N (2002) The use of data envelopment

analysis with probabilistic assurance regions for measuring

hospital efficiency J Prod Anal 17:83–110

6 Tortosa-Ausina E (2002) Bank cost efficiency and output

speci-fication J Prod Anal 18:199–222

7 Sharma K, Leung P (1999) Technical efficiency of the longline

fishery in Hawaii: an application of a stochastic production

frontier Mar Res Econ 13:259–274

8 Kirkley J, Squires D, Strand I (1995) Assessing technical

effi-ciency in commercial fisheries: the Mid-Atlantic sea scallop

fishery Am J Agric Econ 77:686–697

9 Garcı´a del hoyo J, Castilla Espino D, Jime´nez T (2004)

Deter-mination of the technical efficiency of fisheries by stochastic

frontier models: a case on the Gulf of Ca´diz (Spain) ICES J Mar

Sci 61:416–421

10 FAO (2004) Measuring and assessing capacity in fisheries: issues

and methods FAO Fisheries Report 433(2) Rome

11 Kim D (2006) Measurement of fishing capacity of offshore

fisheries in Korea J Fish Bus Admin 37(1):1–24 (in Korean with

English abstract)

12 Kirkley J, Fare R, Grosskopf S, McConnell K, Squires D, Strand I

(2001) Assessing capacity and capacity utilization in fisheries

when data are limited North Am J Fish Manag 21:482–497

13 Castilla Espino D, Garcı´a del hoyo J, Sharp B (2005) Capacity

and capacity utilization of the ‘‘Voracera’’ fleet in the Strait of

Gibraltar Mar Res Econ 20:367–384

14 Tsitsika E, Maravelias C, Wattage P, Haralabous J (2008) Fishing

capacity and capacity utilization of purse seiners using data

envelopment analysis Fish Sci 74:730–735

15 Battese G (1992) Frontier production functions and technical efficiency: a survey of empirical applications in agricultural economics Agric Econ 7:185–208

16 Ferrier G, Lovell C (1990) Measuring cost efficiency in banking: econometric and linear programming evidence J Econ 46:229–245

17 Førsund F, Lovell C, Schmidt P (1980) A survey of frontier production functions and of their relationship to efficiency mea- surement J Econ 13:5–25

18 Sharma K, Leung P, Zaleski H (1997) Productive efficiency of the swine industry in Hawaii: stochastic frontier vs data envel- opment analysis J Prod Anal 8:447–459

19 Charnes A, Cooper W, Rhodes E (1978) Measuring the efficiency

of decision making units Eur J Oper Res 2:429–444

20 Cooper W, Seiford L, Tone K (2007) Data envelopment analysis:

a comprehensive text with models, applications, references and DEA-Solver software Springer, New York

21 Kumbhakar S, Lovell C (2000) Stochastic frontier analysis Cambridge University Press, Cambridge

22 Minh N, Long G, Thang B (2007) Technical efficiency of small and medium manufacturing firms in Vietnam: parametric and non-parametric approach Kor Econ Rev 23:187–221

23 Aigner D, Lovell C, Schmidt P (1977) Formulation and tion of stochastic frontier production function models J Econ 6:21–37

estima-24 Jondrow J, Lovell C, Materov I, Schmidt P (1982) On the mation of technical inefficiency in the stochastic frontier pro- duction function model J Econ 19:239–285

esti-25 Battese G, Coelli T (1988) Prediction of farm-level technical efficiencies with a generalized frontier production function and panel data J Econ 38:387–399

26 Stevenson R (1980) Likelihood functions for generalized chastic frontier estimation J Econ 13:343–366

sto-27 Meeusen W, Van Den Broeck J (1977) Efficiency estimation from Cobb–Douglas production functions with composed error Int Econ Rev 18:435–445

28 Kirkley J, Squires D, Strand I (1998) Characterizing managerial skill and technical efficiency in a fishery J Prod Anal 9:145–160