212 
Fishery Bulletin 11 6(2) 
and Tanaka, 1984) and a female:male ratio of 4.1:1.0 of 
velvet dogfish captured at the Canary Islands (Pajuelo 
et ah, 2010). Segregation may result from the move¬ 
ment of females from depth to epipelagic water because 
a 1:1 sex ratio has been reported for velvet dogfish cap¬ 
tured on bottom longlines at depths as great as 1000 m 
(Kobayashi, 1986; Nakaya and Shirai, 1992; Yano and 
Kugai, 1993). Sexual segregation has been reported for 
other members of the family Somniosidae, including 
the roughskin dogfish (Centroscymnus owstonii) and 
the Portuguese shark (C. coelolepis) (Yano and Tanaka, 
1988; Girard and Du Buit, 1999), both of which are 
relatively large squaloid sharks. 
Females captured in our study were mature or ap¬ 
proaching maturity and therefore indicated that velvet 
dogfish may also segregate by size. Segregation may 
reflect the movement of mature females into epipelagic 
waters, and such movement may be related to repro¬ 
duction rather than to feeding. There is evidence of 
segregation by size for a number of species of dogfish, 
including the closely related Japanese velvet dogfish 
(Scymnodon ichiharai ): large individuals are thought 
to be more common inside the boundary of Suruga Bay, 
Japan, than smaller, immature individuals (Yano and 
Tanaka, 1984). 
Diet 
Our findings indicate that velvet dogfish feed primar¬ 
ily on squid and less on fish and shrimp; however, 
these results may be biased as a result of the longer 
retention of squid beaks in stomachs because of their 
resistance to digestion than the retention of other in¬ 
vertebrate prey or fish (Hyslop, 1980). The stomachs 
and intestines of females captured in the upper water 
column were examined for remains of prey and from an 
examination of these organs, squid appear to form the 
major portion of the diet. It is not surprising that squid 
are a major food item of the velvet dogfish in waters 
surrounding Hawaii, given the abundance of a number 
of species of squid that occur in this region. The great¬ 
est trawl catches for a wide range of cephalopod species 
in Hawaii have been recorded at night in the upper 
200 m (Young, 1978). Many of these cephalopod species 
are thought to undergo a diel vertical migration from 
daytime depths of 400-700 m to shallower nighttime 
depths of 250-300 m (Young, 1978). The well-developed 
tail of velvet dogfish and the depth distribution of the 
various squid consumed by the dogfish examined indi¬ 
cate that the velvet dogfish may be an active feeder or 
an ambush predator in the epipelagic and mesopelagic 
zones. Active predation on pelagic cephalopods by adult 
and subadult Portuguese sharks in the Catalan Sea 
was also reported by Carrasson et al. (1992). 
Female reproductive biology 
Although the number of velvet dogfish specimens ex¬ 
amined in our study was small, the information gained 
from our examination of these sharks expands our un¬ 
derstanding of the reproductive biology of females of 
this species. Yano and Tanaka (1984) reported 3-10 
ova in uteri of female velvet dogfish; however, actual 
brood size may be less in other squaloid sharks (i.e., 
20-31 uterine ova and 16-28 embryos for the roughs¬ 
kin dogfish and 22-29 uterine ova and 15-29 embryos 
for the Portuguese shark) in Suruga Bay, Japan (Yano 
and Tanaka, 1988). The brood size for velvet dogfish 
examined in this study was 2-8 young. Five embryos 
200-220 mm TL have been observed in the last third 
of the gestation period (Romanov et aL, 2013; Gra¬ 
ham 1 ). Size at birth for velvet dogfish is estimated at 
245-270 mm TL on the basis of observations of embry¬ 
os with yolk sac attached at sizes of 200-220 mm TL 
(A. Graham 1 ), free-swimming sharks 250-270 mm TL 
with faint umbilical scars (Graham 1 ; Romanov 2 ), and 
a 277-mm-TL free-swimming shark with an umbilical 
scar completely healed (Graham 1 ). The minimum sizes 
reported for free-swimming individuals were 246-262 
mm TL (Cadenat and Blache, 1981; Yano and Tana¬ 
ka, 1984; Taniuchi and Garrick, 1986; Graham 1 ). This 
information substantially narrows the range for esti¬ 
mates of size at birth from the large size (690 mm TL) 
previously reported by Burgess and Chin (2006) and 
the small size (200 mm TL) reported in Ebert (2015). 
The estimated size at maturity for females was be¬ 
tween 715 and 730 mm TL in our study and was based 
on the size of a 715-mm-TL pregnant female and im¬ 
mature individuals approaching 730 mm TL. These es¬ 
timates reduce the size range at sexual maturity. Our 
estimate of size at maturity for females is substantially 
larger than the estimate of 590 mm TL reported by 
Compagno et al. (2005). Our current hypothesis is that 
there is no annual seasonal cycle of reproduction for 
this species; however, more data are required to con¬ 
firm this notion. In waters off Hawaii, ovulating fe¬ 
males and early-stage gravid females (small embryos 
31.1-60.4 mm) were observed between April and Au¬ 
gust. Lack of a seasonal reproductive cycle is a common 
feature of squaloid sharks (Yano and Tanaka, 1984). 
The entirety of data available on reproduction in fe¬ 
male velvet dogfish provides some evidence that they 
may be associated with the benthic habitat for parturi¬ 
tion because of the capture of small, free-swimming in¬ 
dividuals and females with near-term embryos in utero 
on bottom longlines (Romanov et al., 2013). 
The velvet dogfish is an enigmatic species captured 
across a wide geographical range and throughout a 
range of depths in the water column. However, this 
species is rare in pelagic fisheries, and therefore it is 
typically caught in low numbers in fisheries around the 
world. For example, Dai et al. (2006) reported a catch 
per unit of effort (CPUE) for velvet dogfish of only 
0.007 individuals/1000 hooks for longline fisheries in 
the eastern Pacific Ocean. In the Hawaii longline fish¬ 
eries (swordfish and tuna longlines combined) during 
1995-2006, the CPUE of velvet dogfish was 0.005/1000 
hooks (Walsh et al., 2009). In the eastern central Pa¬ 
cific Ocean, velvet dogfish were caught primarily at a 
median depth of 190 m, and the box plot of 25 and 
