Marine Biology (2000) 137: 519±526

Ó Springer-Verlag 2000

B. A. Seibel á F. G. Hochberg á D. B. Carlini

Life history of Gonatus onyx (Cephalopoda: Teuthoidea): deep-sea spawning and post-spawning egg care

Received: 22 February 1999 / Accepted: 25 May 2000

Abstract A reproductive strategy consisting of deepwater spawning and egg-care was inferred for the midwater squid Gonatus onyx Young, 1972. Brooding females and associated eggs and hatchlings, captured between 1250 and 1750 m depth o€ southern California, are described. Brooding females appear to be senescent and lack tentacles. Large eggs of this species (3 mm) at cold temperatures (3 °C at capture depth) may require as long as 9 mo to develop. The high lipid content of the digestive gland in adult females of this species may provide fuel to support such an extended ``brooding'' period.

Introduction The life-history strategies of oceanic cephalopods are largely unknown. Based primarily on knowledge of a few shallow-living and commercially important squids (order Teuthoidea), it has been assumed that postspawning egg-care is limited to the order Octopoda (Young 1972; von Boletzky 1992). As a result, eggbrooding and related characteristics have been used to

derive evolutionary relationships between the squids and octopods (von Boletzky 1992). However, two squid photographed in the Sea of Okhotsk carrying large egg masses were recently reported (Okutani et al. 1995). Although the specimens were not captured, they were, based on the absence of tentacles, tentatively identi®ed as squids in the genus Gonatopsis (Gonatidae). Gonatus onyx Young, 1972 is among the most abundant cephalopods o€ the coast of California (Roper and Young 1975). Because of an ontogenetic descent to great depths (Hunt and Seibel 2000), spawning, egg masses, eggs and hatchlings have never been observed for this species. The present study provides the ®rst report of brooding behavior in the mesopelagic squid, G. onyx. Two senescent females, eggs (in addition to fragments of the egg mass) and hatchlings captured between 1250 and 1750 m o€ southern California are described. The energetic and evolutionary advantages of the observed life-history characteristics are discussed.

Materials and methods

Communicated by N. H. Marcus, Tallahassee

Collection

B. A. Seibel (&) á F. G. Hochberg á D. B. Carlini Rosenstiel School of Marine and Atmospheric Science, Division of Marine Biology and Fisheries, 4600 Rickenbacker Causeway, Miami, Florida 33149-1098, USA

Specimens of Gonatus onyx Young, 1972 were captured in an opening±closing Mother Tucker trawl equipped with a thermallyprotecting cod-end (Childress et al. 1978). BESS (biological environmental sampling system) pressure and salinity and SEA BIRD temperature sensors, which relay data to a computer-controlled shipboard acquisition controller via a single-conductor armored cable, were deployed with the net. Specimens were captured 160 km west of Point Conception, California (35°0¢N; 123°02¢W) and over the San Clemente Basin o€ San Diego, California (32°27¢W; 117°57¢W). Live specimens were weighed on a motion-compensated, precision, ship-board balance system (Childress and Mickel 1980). Tissue samples were taken and frozen in liquid nitrogen for gene-sequencing and biochemical analysis. Specimens were then preserved in 5% formalin in seawater, and are presently archived in the collections of the Santa Barbara Museum of Natural History, Santa Barbara, California (SBMNH). Morphometric and meristic measurements were made according to Roper and Voss (1983) on preserved specimens and eggs.

Fax: 001 (305) 361-4600 e-mail: [email protected] Present addresses: F. G. Hochberg Department of Invertebrate Zoology, Santa Barbara Museum of Natural History, Santa Barbara, California 93105-2936, USA D. B. Carlini Department of Biology, University of Rochester, Rochester, New York 14627-0211, USA

520 Gene-sequencing Nucleotide sequence of the cytochrome-C oxidase (CCO) gene was determined for Gonatus onyx and additional gonatid species following the methods described in Carlini and Graves (1999). The LCO 1490 and HCO 2198 polymerase chain-reaction (PCR) primers designed by Folmer et al. (1994) were used to amplify the cytochrome oxidase I (COI) gene in gonatids. DNA sequences were aligned by eye and compiled in MacClade 3.0 (Maddison and Maddison 1992). Alignment was straightforward, as there were no apparent insertion/deletion events or alignment ambiguities, a ®nding consistent with the results of Folmer et al., whereby no gaps were introduced in the alignment of COI sequences from diverse metazoan phyla. Non-gonatid COI sequences obtained by Carlini and Graves were included in the analyses, and were used as outgroup sequences. Phylogenetic analysis of the aligned sequences (657 bp, base excluding the primer sequences) was conducted using the branch-and-bound search option in PAUP* (phylogenetic analysis using parsimony: Swo€ord 1993). Both parsimony and maximum-likelihood were implemented as the optimality criteria in tree searches. Maximum-likelihood analysis was conducted using the nucleotide substitution model of Hasegawa et al. (1985), with gamma-distributed rates partitioned into four discrete categories Fig. 1 Gonatus onyx. A Ventro-lateral view of two mature female specimens; Specimen No. 1 (right, SBMNH 345288) has a mantle length (ML) of 132 mm; this specimen was captured with an egg mass (B). Specimen No. 2 (left, SBMNH 345289) has a ML of 145 mm; this specimen was captured with hatchling squids (C); The digestive gland (arrowed) of specimen No. 2 is smaller and degenerate relative to that of specimen No. 1. B fragment of an egg mass containing some eggs. C Hatchlings (preserved in formalin) with a ML of 3.4 mm

(Yang 1994). Support for phylogenetic trees was examined using the branch-and-bound bootstrap search command (1000 replicates) in PAUP*, which was also used to calculate uncorrected sequence divergences for pairwise comparisons among taxa.

Results Material examined Brooding females (Fig. 1a; Table 1) Two mature female Gonatus onyx were captured o€ California. Specimen No 1 (SBMNH 345288) was collected on 21 September 1995 in an oblique tow between 1250 and 1750 m o€ Point. Conception, California, over a bottom depth of '4100 m. It had a mantle length (ML) of 132 mm and a wet weight of 99.7 g. Associated with this specimen were eggs embedded in a fragile black gelatinous material. Specimen No. 2 (SBMNH 345289) was collected on 22 May 1996 between 1350 and 1420 m depth in the San Clemente Basin o€ San Diego, California, over a bottom depth of 2100 m. It had a mantle length of 145 mm and a wet weight of 138.6 g. Associated with this specimen were '2000 hatchlings.

521 Table 1 Gonatus onyx. Morphometric and meristic measurements of two brooding females; Santa Barbara Museum of Natural History Catalogue Nos. 345288 (Specimen No. 1) and 345289 (Specimen No. 2). Length units are mm and weights are g (*estimate, mantle cut; **estimate, arm cut; C cut; L left; R right; L-AL locking-apparatus length, (n nuchal, f funnel; ± no data) Characteristic

Specimen No. 1

Specimen No. 2

Maturity Total wet weight Dorsal mantle length Ventral mantle length Mantle width Head width Funnel length Funnel organ length (medial) Funnel organ length (lateral) L-AL (n) Eye diam Fin length Fin width L-AL (f) Gill length Gill lamellae number

With eggs With hatchlings 99.68* 138.6* 132 145 131* 135* 35* 38* 32 33 20 21 17 16 11 11 16 15 15.7 L 16.8 R 11.4 L 11.8 R C 52.7 62.5 C C 41.5 52.0 C 21.6 20.8 20.0 20.0 47.9 49.7 48.5 49.2 56 54 52 55

Arm length 1 2 3 4 Arm length formula

62 70 72 C 3.2 = 4.1

57 68 72 69

100** 110 118 106 3 = 2.4.1

107 120 115 112

Hook No. 1 2 3 4

31 31 32 ±

33 30 34 ±

30** 30 33 ±

31 33 30 ±

Depth (m)

1250±1755

Egg mass and eggs associated with female No. 1 (Figs. 1B, 2A±C) A number of egg-mass fragments, associated with Female No. 1, were examined. The eggs were embedded in soft, fragile, opaque jelly-like material (Figs. 1B, 2A, B), and appear to be deposited in a ¯at sheet a single egg thick. The egg mass is composed of two thin membranes that form individual chambers in a honeycomb-like pattern. Membranes are fused between each individual chamber, which helps to bind the egg mass together and hold the individual eggs in place (Fig. 2B). Eggs are oriented at random in the egg mass. In any given fragment of the gelatinous matrix, eggs are present in the majority of the individual chambers. In some cases, when the surrounding membranes have been torn, the egg chambers are empty. The egg mass is black. It is darkest where the membranes are fused between each egg chamber, and lighter black or greyish black where the membrane thins over each egg. The cover of each egg chamber is ¯ecked with spots of dark pigment (Fig. 2A). The color may derive from the ink of the female squid. Preserved eggs are ovoid or elliptical in shape. Lengths range from 2.0 to 3.0 mm, widths from 1.8 to 2.1 mm. The chorion capsule wall is thin throughout. All embryos are at the same early stage of development. The eggs are predominantly a light yellowish color, presumably derived from the color of the yolk material present. A distinct white band is present at one end of each egg (Fig. 2C).

1350±1420

The specimen associated with hatchlings (Specimen No. 2) was senescent and had lower enzymatic activities (Hunt and Seibel 2000) than the specimen associated with undeveloped eggs (Specimen No. 1). The digestive gland of Specimen No. 1 was ®rm, while that of Specimen No. 2 was ¯accid and watery (Fig. 1A). The beak in each specimen was pigmented deep red. Upon capture, remnants of black material of a consistency indistinguishable from that of the egg mass were articulated around the hooks and suckers on the arms of both mature female specimens. The present specimens are larger than any previously reported specimens of this species. Specimen No. 1 still had some oocytes in the ovary. The ovary of Specimen No. 2 was completely spent. Based on molecular data (see subsection ``Molecular phylogenetic analyses'' below), we are con®dent in our identi®cation of the present specimens as Gonatus onyx. However, because tentacles, absent from the present specimens, have been most widely used in the identi®cation of Gonatus species, and because the mature female specimens presently reported are larger than any previously reported specimens of this species, morphometric and meristic measurements are provided in Table 1. Photographs and drawings of a fragment of the egg mass, hatchlings and adults also are provided (Figs. 1, 2).

Hatchlings associated with female No. 2 (Figs. 1C, 2D±I) The hatchlings are moderately large. Total lengths of preserved specimens are up to 5.0 mm, and total wet weights average 0.01 g. Mantle lengths range from 3.2 to 3.5 mm and mantle widths are 50 to 60% of ML. The mantle is elongate, cylindrical or bell-shaped. The mantle wall is thin and muscular. The hatching gland is evident as a glandular swelling on the posterior dorsal mantle. Fins are minute; ®n lengths are 4 to 5% of ML, and ®n widths 15 to 20% of ML. The head often is contracted in the mantle. The eyes are moderately large, 15% of ML. The arms are short and relatively stout. The arm formula (see Roper and Voss 1983 for de®nition) is 1 = 2.4 = 3; Arms 1 are 18 to 20% of ML; Arm Pair 1 have two stalked suckers, Arm Pair 2 have two stalked suckers + 1 sucker bud, Arm Pairs 3 and 4 lack both hooks and suckers. The tentacles are 2 to 3 times longer than the arms, and on average are 40% of ML. The tentacles are thick and muscular, and a terminal ®lament is present. The oral surface of the tentacle club is ¯attened and densely packed with 70 to 90 minute sucker buds. The central hook and ®xing apparatus are not visible. A characteristic chromatophore pattern is present at the time of hatching (Fig. 2D±F), namely 5 to 6 in a single row on the aboral surface of each tentacle, a single pair at

522

the base of the ®ns on the posterior tip of dorsal mantle, 1 at the anterior end of the hatching gland on the middorsal mantle, and 2 to 3 pairs on the lateral mantle. The hatchlings observed with Female No. 2 are by far the smallest and youngest paralarval stages described to date for any gonatid squid (Arkhipkin and Seibel 1999). No remnant yolk stores are apparent. Hatchlings were observed live on board ship in a small (25 ´ 25 ´ 5 cm) aquarium maintained at 5 °C. They appeared to be negatively buoyant and swam with the hop-and-sink swimming style described by O'Dor Fig. 2 Gonatus onyx. A±C Egg cluster associated with female specimen No. 1 (SBMNH 345288): A Fragment of egg mass with eggs arranged in single layer: B Side view of egg in cluster; C Detail of egg. D±F Hatchlings associated with female specimen No. 2 (SBMNH 345289): D Dorsal view; E lateral view of right side of hatchling; F Ventral view. G±H Details of arm crown of hatchling to show arrangement of arms and tentacles: G Oral view with tentacles and ventral arms removed; H Lateral view; I details of contracted tentacle (left oral view; right lateral view). [A1±A4 arm pair numbers (e.g., A1 = dorsal arm pair 1); BM buccal region; EC empty egg chamber; F ®n; Fu funnel; GM gelatinous matrix; IS ink sac; HG hatching gland; T tentacles]

et al. (1986). Many hatchlings eventually settled to the bottom and only swam when disturbed. Molecular phylogenetic analyses Nucleotide-sequence divergences from pairwise comparisons among gonatid COI sequences (Table 2) indicate that di€erences between the present specimens (including hatchlings) and voucher specimens of Gonatus onyx range from 0.15 to 1.37% (1 to 9 bases out of 657) or less, whereas comparisons between Gonatus congen-

523

ers range from 8.38 to 9.89% (55 to 65 bases out of 657). Di€erences between Gonatopsis borealis and species in the genus Gonatus ranged from 11.3 to 13.87%, a divergence consistent with previous estimates of COI divergences within oegopsid families (Carlini and Graves 1999). Gonatid species di€ered from Thysanoteuthis rhombus, Loligo pealei, and Sepioteuthis australis by 17 to 20%, a range comparable to the average divergence between non-confamilial teuthoid species selected at random (Carlini and Graves 1999). A phylogenetic tree constructed from parsimony analysis of the COI data strongly supports the identi®cation of brooding females and four hatchlings as Gonatus onyx, as indicated by the 100% bootstrap support for the clade containing the present specimens and voucher G. onyx specimen. Maximum-likelihood analysis conducted on the nucleotide-sequence data yielded a tree topology identical to the parsimony tree, and bootstrap analysis also strongly supported the monophyly of the G. onyx + brooding females + hatchlings clade (results not shown).

Discussion Clarke (1966) pointed out >30 yr ago that ``our almost complete ignorance of the eggs and early larval stages of oegopsid squids is the most extraordinary omission in our knowledge of cephalopod biology''. While our knowledge of the larvae of oceanic cephalopods has improved considerably since then (for review see Sweeney et al. 1992), with few exceptions, our knowledge of the eggs (O'Dor and Balch 1985; Young et al. 1985a, b; Okutani et al. 1995; Bjorke et al. 1997), has not. The recent photographs by Okutani et al. (1995) document, for a gonatid squid, a novel reproductive strategy within the order Teuthoidea. Subsequent work by Bjorke et al. (1997) suggests the possibility of a similar strategy for Gonatus fabricii (Lichtenstein, 1818). The present observations on G. onyx substantiate the previously observed brooding habit among gonatid squid species.

Gonatid phylogenetic relationships Parsimony analysis and maximum-likelihood analysis strongly supported the monophyly of the Gonatidae, the monophyly of the genus Gonatus, and the monophyly of the G. onyx clade (Fig. 3). However, relationships within the genus Gonatus were not strongly supported, as indicated by the lack of bootstrap support for the (G. californiensis±G. berryi) and (G. fabricii±G. onyx) clades. Sequences from the remaining seven species in the genus, as well as additional representatives of the genus Gonatopsis and representatives of the genera Eogonatus and Berryteuthis, are required to obtain a more robust phylogeny of the family. It does appear that the COI gene is suciently variable within the family to be useful in elucidating gonatid relationships. Tentacle loss The absence of tentacles in the brooding specimens originally led us to attribute these specimens to the genus Gonatopsis. However, most gonatid species lose their tentacles upon maturation or spawning (Young 1973; Kristensen 1981; Okiyama 1993). Among gonatids, only Gonatopsis loses tentacles earlier in its life cycle (Okiyama 1969, 1993). We suggest that tentacle loss facilitates holding of an egg mass in the arms. Tentacles are used primarily for feeding and, in some cases, mating. Therefore, tentacle loss that occurs after cessation of feeding (usually just after spawning) should have no e€ect on ®tness. Buoyancy, energetics and ontogeny Many gonatid squids are believed to be neutrally buoyant based on high levels of lipid (12 to 20% body lipid composition), especially diacyl glyceryl ethers (DAGE), concentrated in the digestive gland (Clarke et al. 1979; Kristensen 1984; Hawashi and Kawasaki 1985; Hayashi and Yamamoto 1987; Hawashi 1989).

Table 2 Gonatidae. Pairwise distances between cytochrome C oxidase I sequences [Above diagonal % sequence divergence (= total number of di€erences ¸ 657 nucleotides ´ 100%); below

diagonal total number of nucleotides that di€er between the two species compared; undelined numbers de®ne all conspeci®c pairwise comparisons (i.e. only those within Gonatus onyx)]

Species 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1

Gonatus onyx 8 Brooding Female No. 1 (w/eggs) Brooding Female No. 2(w/hatchlings) 3 2 Hatchling #1 Hatchling #2 2 4 Hatchling #3 Hatchling #4 1 Gonatus berryi 61 Gonatus californiensis 57 Gonatus fabricii 59 Gonatopsis borealis 84 Thysanoteuthis rhombus 114 Loligo pealei 143 Sepioteuthis australis 136

2

3

4

5

6

7

8

9

1:22

0:46 1:37

0:30 1:22 0:46

0:30 1:22 0:15 0:30

0:61 1:37 0:76 0:61 0:61

0:15 1:07 0:30 0:15 0:15 0:46

9.29 9.74 9.59 9.59 9.44 9.89 9.44

8.69 8.38 8.69 8.54 8.54 8.99 8.54 7.93

9 8 8 9 7 64 55 60 83 112 143 134

3 1 5 2 63 57 60 85 113 144 135

2 4 1 63 56 59 84 114 145 136

4 1 62 56 59 84 113 144 135

3 65 59 61 86 112 147 136

62 56 58 84 113 144 135

52 64 91 120 147 143

47 74 108 136 129

10 8.98 9.13 9.13 8.98 8.98 9.29 8.83 9.74 7.17 77 122 147 126

11

12

13

14

12.81 12.65 12.96 12.81 12.81 13.11 12.81 13.87 11.3 11.74

17.35 17.05 17.20 17.35 17.20 17.05 17.20 18.27 16.46 18.57 17.23

21.77 21.77 21.92 22.07 21.92 22.37 21.92 22.37 20.73 22.73 22.26 19.18

20.70 20.40 20.55 20.7 20.55 20.70 20.55 21.77 19.67 19.18 20.88 19.64 17.50

113 146 137

126 129

115

524

Fig. 3 Gonatidae. Most parsimonious tree derived from a branchand-bound parsimony search of the 657 base-pair fragment of cytochrome C oxidase I gene implemented in software program PAUP* 4.0 (Swo€ord 1998). Tree length = 446 steps; consistency Index = 0.709; retention index 0.629 [Numbers above the branches number of unambigous changes occurring along branch; boxed numbers to left of nodes percentage of 1000 bootstrap replicates that support the node (values <50% not shown)]. Branch lengths are proportional to amount of change. Maximum-likelihood analysis of data yielded identical tree topology. GenBank Accession Nos. for COI sequences are: Gonatus onyx: AF000041; Brooding Female No. 1: AF144718; Brooding Female No. 2: AF144719; Hatchling No. 1: AF144720; Hatchling No. 2: AF144721; Hatchling No. 3: AF144722; Hatchling No. 4: AF144723; Gonatus berryi: AF000040; Gonatus californiensis: AF144724; Gonatus fabricii: AF131873; Gonatopsis borealis: AF144752; Thysanoteuthis rhombus: AF000070; Loligo pealei: AF000052; Sepioteuthis australis: AF000065

Rapid synthesis of DAGE is responsible for regulation of buoyancy in the dog®sh Squalus acanthias (Malins and Barone 1970). However, buoyancy adaptations can be detrimental to locomotion, and are not likely to be strongly selected for in active animals such as Gonatus onyx (O'Dor 1988; Seibel et al. 1997; Hunt and Seibel 2000). Clarke et al. (1985) reported that early life stages of Gonatus spp. are, in fact, low in lipid content and negatively buoyant. The hatchling G. onyx described here appeared to be negatively buoyant and utilized a hop-and-sink swimming style (personal observations) as seen in paralarvae of some species of muscular squid (O'Dor et al. 1986). Buoyancy in adult individuals may counter the increased weight of eggs and reproductive organs (Voight et al. 1994). Buoyancy may be essential in pelagic animals whose locomotory abilities are restricted by the burden of carrying egg masses. High lipid content in older individuals may not (only) be used for buoyancy, but (also) as an energy reserve utilized during an extended brooding period. Feeding does not occur in the midwater octopod Eledonella pygmaea during egg-brooding (Young 1972), and seems

unlikely to occur during egg-brooding in any midwater cephalopod because of the preoccupation of the arms with handling the egg mass. With a 12 to 20% body lipid content, a metabolic rate of 2.92 ml kg)1 min)1 (scaled to 100 g at 3 °C: Seibel et al. 1997), an energy content of 9.4 kcal g)1 lipid, and an oxycalori®c conversion of 4.7 kcal l)1 O2 (O'Dor and Webber 1986), Gonatus onyx could survive as long as 120 d using lipid as its only fuel source. Decreased metabolism in deep-living, brooding specimens (Hunt and Seibel 2000) could extend this estimate more than three times. The use of lipids to fuel an extended spawning period is consistent with the more degenerate condition of the digestive gland of Female No. 2 (associated with hatchlings) compared to Female No. 1 (associated with undeveloped eggs: Fig. 1A). The lipid content of male adult G. onyx is not known. Kristensen (1984) suggested that male G. fabricii live ³2 yr, and perhaps spawn more than once. Hunt and Seibel (2000) reported two year-classes of G. onyx present during summer months in Monterey Bay, suggesting at least a 2 yr life cycle (not including the brooding period in females). There may be sexual dimorphism in body size, metabolism, behavior and/or lipid content of the digestive gland as well as sexual vertical nichedivergence, as has been suggested for another brooding, midwater cephalopod, Eledonella pygmaea (Voight 1995). These hypotheses await further testing. Brooding period Spawning in Gonatus onyx occurs from April to June or July in the California Current (Okutani and McGowan 1969; Okutani et al. 1988). This is consistent with the capture of Specimen No. 2, (associated with recent hatchlings) in late May. In the deep waters where the present specimens were captured, temperatures near

525

3 °C could prolong development of the embryos (based on an egg size of 3 mm) for ³9 mo (present Fig. 4; von Boletzky 1994; Wood et al. 1998)! Given that spawning is seasonal and that hatching occurs only in summer, Female No. 1 (captured in September with undeveloped eggs) was just beginning and Female No. 2, (captured in May with hatchlings) just ending what may be a 9 mo brooding period. This is within the survival range estimated from lipid reserves. Life-history evolution Large eggs and extended development times may be an adaptation which allows production of more advanced paralarvae capable of a long vertical migration to nutrient-rich shallower waters (Mangold 1987). However, Okutani et al. (1995) observed gonatid hatchlings being released in shallow water. Brooding females of gonatid squid species may migrate to the surface to release hatchlings. The hatchlings captured in the present study were perhaps released prematurely in response to the trauma of net capture, as happens in brooding midwater octopods (Young 1972). Swimming activity for hatchlings on board ship was extremely low at low temperatures (Seibel personal observations); this may be partially related to premature hatching. A migration of >1200 m to the surface seems unlikely for these hatchlings, especially given their apparent negative buoyancy, the absence of remaining yolk stores, cold temperatures, and the low food availability at depth.

Increased parental investment may be selected for by the stability of the deep-sea environment (Childress et al. 1980), where parental survival is high due to decreased predation pressure. The epipelagic habitat and small sizes of most cephalopod paralarvae, on the other hand, presumably result in high juvenile mortality. Large eggs, requiring longer development times, would presumably also be at greater risk of predation in epipelagic waters. Where parental survival is high and that of the juveniles is low, selection will tend to favor species that put energy into creating larger more capable o€ spring (Stearns 1992; Wood et al. 1998). The great paucity of data on spawning habits of midwater cephalopods leads us to speculate that many species are spawning at great depths. Deep-sea spawning has, in fact, been observed for several mesopelagic species, including other gonatids, cranchiids, and octopods (Young 1972, 1975; Bjorke et al. 1997). Large eggs, long development times, and post-spawning parental care may be common strategies. Increased ®shing at great depths is required to elucidate the life-history strategies of midwater cephalopods. Acknowledgements We thank J. J. Childress, J. C. Hunt and J. Company for helpful discussions and constructive comments on the manuscript. Illustrations of eggs and hatchlings (Fig. 2) were prepared by J. T. Rounds. We are grateful to the captains and crews of the research vessels ``New Horizon'', ``Pt. Sur'' and ``David Starr Jordan''. We would also like to thank D. Woodbury for providing some samples. An NSF Doctoral Dissertation Improvement Grant (DEB-9623353) awarded to D. B. Carlini and J. E. Graves supported the molecular aspect of the study. Other aspects of this research were supported by NSF Grant OCE 9415543 to J. J. Childress.

References

Fig. 4 Cephalopods. Relationship between egg length (x = mm) and development time (y = days) for a variety of cephalopod species (d) measured or estimated between 5 and 7 °C (y = 70.67x0.86; R2 = 0.70). Gonatus onyx (´), with an egg length of 3 mm, is shown at its estimated development time of 270 d at 3 °C [1, Illex illecebrosus; 2, Loligo pealei: 3, L. vulgaris; 4, Sepietta oweniana; 5, Rossia paci®ca; 6, Octopus do¯eini.(data for species 1 to 6 are summarized by Boletzky (1994). 7, Bathypolypus arcticus (Wood et al. 1998)]. Data for Species 1 to 4 are extrapolations outside natural range of temperatures experiencd by these species

Arkhipkin AI, Seibel BA (1999) Statolith microstructure from hatchlings of the oceanic squid, Gonatus onyx (Cephalopoda: Gonatidae) from the Northeast Paci®c. J Plankton Res 21: 401±404 Bjorke H, Hansen K, Sundt R (1997) Egg masses of the squid Gonatus fabricii (Cephalopoda, Gonataidae) caught with pelagic trawl o€ northern Norway. Sarsia 82: 149±152 Boletzky S von (1992) Evolutionary aspects of development, life style, and reproductive mode in incirrate octopods (Mollusca: Cephalopoda). Revue suisse Zool 99: 755±770 Boletzky S von (1994) Embryonic development of cephalopods at low temperatures. Antarctic Sci 6: 139±142 Carlini DB, Graves JE (1999) Phylogenetic analysis of cytochrome c oxidase I sequences to determine higher-level relationships within the coleoid cephalopods. Bull mar Sci 64: 57±76 Childress JJ, Barnes AT, Quetin LB, Robison BH (1978) Thermally protecting cod ends for recovery of living deep-sea animals. Deep-Sea Res 25A: 419±422 Childress JJ, Mickel TJ (1980) A motion compensated ship-board precision balance system. Deep-Sea Res 27A: 965±970 Childress JJ, Taylor SM, Cailliet GM, Price MH (1980) Patterns of growth, energy utilization and reproduction in some mesoand bathypelagic ®shes o€ southern California. Mar Biol 61: 27±40 Clarke A, Clarke MR, Holmes LJ, Waters TD (1985) Calori®c values and elemental analysis of eleven species of oceanic squids (Mollusca: Cephalopoda). J mar biol Ass UK 65: 983±986

526 Clarke MR (1966) A review of the systematics and ecology of oceanic squids. Adv mar Biol 4: 91±300 Clarke MR, Denton EJ, Gilpin-Brown JB (1979) On the use of ammonium for buoyancy in squids. J mar biol Ass UK 59: 259±276 Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primers for ampli®cation of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molec mar Biol Biotechnol 3: 294±299 Hasegawa M, Kishino H, Yano T (1985) Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J molec Evolut 22: 160±174 Hawashi K (1989) Occurrence of diacyl glyceryl ethers in liver lipids of gonatid squid, Gonatopsis borealis. Nippon Suisan Gakk 55: 1383±1387 Hawashi K, Kawasaki K (1985) Unusual occurrence of diacyle glyceryl ethers in liver lipids from two species of gonatid squids. Bull Jap Soc scient Fish 51: 593±597 Hayashi K, Yamamoto S (1987) Distribution of diacyl glyceryl ethers in the di€erent tissues and stomach contents of gonatid squid Berryteuthis magister. Nippon Suisan Gakk 53: 1057±1063 Hunt JC, Seibel BA (2000) Life history of Gonatus onyx (Cephalopoda): ontogenetic changes in habitat, behavior and physiology. Mar Biol 136: 543±552 Kristensen TK (1981) First record of a mature female of the squid Gonatus fabricii (Lichtenstein, 1818) (Cephalopoda: Teuthoidea). Steenstrupia 7: 101±108 Kristensen TK (1984) Biology of the squid Gonatus fabricii (Lichtenstein, 1818) fom West Greeland waters. Meddr Grùnland (Biosci) 13: 3±17 Maddison WP, Maddison DR (1992) MacClade: Analysis of phylogeny and character evolution. Sinauer Associates, Sunderland, Massachusetts Malins DC, Barone A (1970) Glyceryl ether metabolism: regulation of buoyancy in dog®sh Squalus acanthias. Science, NY 167: 79±80 Mangold K (1987) Reproduction. Vol 2 In: Boyle PR (ed) Cephalopod life cycles. Academic Press, London, pp 157±200 O'Dor RK (1988) The forces acting on swimming squid. J exp Biol 137: 421±442 O'Dor RK, Balch N (1985) Properties of Illex illecebrosus egg masses potentially in¯uencing larval oceanographic distribution. NW Atlant Fish Orgn (NAFO) scient Coun Stud 9: 69±76 O'Dor RK, Foy EA, Helm PL, Balch N (1986) The locomotion and energetics of hatchling squid, Illex illecebrosus. Am malac Bull 4: 55±60 O'Dor RK, Webber DM (1986) The constraints on cephalopods: why squid aren't ®sh. Can J Zool 64: 1591±1605 Okiyama M (1969) A new species of Gonatopsis from the Japan Sea, with the record of a specimen referable to Gonatopsis sp Okutani 1967 (Cephalopoda: Oegopsida, Gonatidae). Publs Seto mar biol Lab 17: 19±32 Okiyama M (1993) Why do gonatid squid Berryteuthis magister lose tentacles on maturation? Nippon Suisan Gakk 59: 61±65

Okutani T, Kubodera T, Je€erts K (1988) Diversity, distribution and ecology of gonatid squids in the subarctic Paci®c: a review. Bull Ocean Res Inst Univ, Tokyo 26: 159±192 Okutani T, McGowan JA (1969) Systematics, distribution, and abundance of the epiplanktonic squid (Cephalopoda, Decapoda) larvae of the California Current April, 1954 ± March, 1957. Bull Scripps Instn Oceangr (New Ser) 14: 26±31 Okutani T, Nakamura I, Seki K (1995) An unusual egg-brooding behavior of an oceanic squid in the Okhotsk Sea. Venus, Kyoto 54: 237±239 Roper CFE, Voss GL (1983) Guidelines for taxonomic descriptions of cephalopod species. Mem natn Mus Vict 44: 49±63 Roper CFE, Young RE (1975) Vertical distribution of pelagic cephalopods. Smithson Contr Zool 209: 1±51 Seibel BA, Thuesen EV, Childress JJ, Gorodezky LA (1997) Decline in pelagic cephalopod metabolism with habitat depth re¯ects di€erences in locomotory eciency. Biol Bull mar biol Lab, Woods Hole 192: 262±278 Stearns SC (1992) The evolution of life histories. Oxford University Press, Oxford Sweeney MJ, Roper CFE, Mangold KM, Clarke MR, Boletzky Sv (1992) ``Larval'' and juvenile cephalopods: a manual for their identi®cation. Smithson Contr Zool 513: 1±282 Swo€ord DL (1998) PAUP*: (phylogenetic analysis using parsimony. Version 3.1.1. Illinois Natural History Museum, Champaign, Illinois Voight JR (1995) Sexual dimorphism and niche divergence in a midwater octopod (Cephalopoda: Bolitaenidae). Biol Bull mar biol Lab, Woods Hole 189: 113±119 Voight JR, Portner HO, O'Dor RK (1994) A review of ammoniamediated buoyancy in squids (Cephalopoda: Teuthoidea). Mar Freshwat Behav Physiol 25: 193±203 Wood JB, Kenchington E, O'Dor RK (1998) Reproduction and embryonic development time of Bathypolypus arcticus, a deep-sea octopod (Cephalopoda: Octopoda). Malacologia 39: 11±19 Yang Z (1994) Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J molec Evolut 39: 306±314 Young RE (1972) Brooding in a bathypelagic octopus. Pacif Sci 26: 400±403 Young RE (1973) Evidence for spawning by Gonatus sp. (Cephalopoda: Teuthoidea) in the high arctic ocean. Nautilus 87: 53±58 Young RE (1975) Leachia paci®ca (Cephalopoda, Teuthoidea): spawning habitat and function of the brachial photophores. Pacif Sci 29: 19±25 Young RE, Harman RF, Mangold KM (1985a) The eggs and larvae of Brachioteuthis sp. (Cephalopoda: Teuthoidea) from Hawaiian waters. Vie Milieu 35: 203±209 Young RE, Harman RF, Mangold KM (1985b) The common occurrence of oegopsid squid eggs in near-surface oceanic waters. Pacif Sci 39: 359±366

Note added at proof

an extended brooding period. Arkhipkin and Bjorke suggest that the gelatinization of the musculature provides buoyancy to support the egg mass and that females are unable to actually protect their eggs. However, the high enzymatic activity (Hunt and Seibel 2000) in Specimen 1 (with undeveloped eggs), relative to Specimen 2 (with hatchlings) suggests that the muscle degeneration is a gradual process and that brooding females may be capable of defending their eggs throughout much, if not all, of the brooding period. In support of our hypothesis that tentacle loss facilitates egg mass brooding, Arkhipkin and Bjorke also observed that tentacles were lost in all spent females but retained by mature males.

Arkhipkin and Bjorke (1999) recently examined the ontogenetic changes in morphometrics and reproductive indices that occur in Gonatus fabricii. Like G. onyx (Hunt and Seibel 2000), G. fabricii apparently accumulates large lipid stores and descends to great depths as adults where spawning takes place. They were able to examine both males and females and reported that both sexes accumulate large lipid stores. The slight decrease in digestive gland weight in mature, relative to maturing, males, suggests that some of the lipid is used to fuel production of spermatophores. However, males still retain substantial lipid stores that may serve a buoyancy funtion and may be retained for additional reproductive events in following seasons. In contrast, no decrease in digestive gland weight was observed for females with development of the reproductive system. However, a dramatic decrease in digestive gland weight was observed in spent females. They also observed a degeneration of the body musculature in both mated and spent females. These observations are consistent with our hypothesis that lipid and protein are used to fuel

Arkhipkin AI, Bjorke H (1999) Ontogenetic changes in morphometric and reproductive indices of the squid Gonatus fabricii (Oegopsida, Gonatidae) in the Norwegian Sea. Polar Biol 22: 357±365 Hunt JC, Seibel BA (2000) Life history of Gonatus onyx (Cephalopoda: Teuthoidea): ontogenetic changes in habitat, behavior and physiology Mar Biol 136: 543±552