Holder and Field: Factors that relate to the occurrence of multiple brooding in Sebastes spp. 
181 
can result in bias and misspecification of stock status and 
associated parameters. For example, if size-dependent 
fecundity cannot be quantified in an assessment but 
such a relationship exists in a real population, assess¬ 
ment models will be biased toward estimating a stock 
status that is more optimistic than actually exists (He 
et aL, 2015a; Barneche et aL, 2018). There has been a 
growing awareness in recent years that the biological 
processes that determine a population’s productivity are 
much more complicated than previously understood. For 
example, some Sebastes species exhibit skipped spawn¬ 
ing (Rideout et aL, 2005; Conrath, 2017), prolonged 
adolescent periods (Thompson and Hannah, 2010), and 
multiple brooding (Moser, 1967; Love et aL, 1990; Beyer 
et aL, 2015; Lefebvre et aL, 2018). Accounting for all of 
these factors is critical in the understanding of a species’ 
life Mstory and in specifying life history characteristics 
in stock assessment models, climate vulnerability assess¬ 
ments, and other evaluations. 
Previous studies indicate that multiple brooding more 
commonly occurs in southerly distributed rockfish spe¬ 
cies (32~36°N), as well as in individuals residing in 
the southern extent of the range of their species (Love 
et aL, 1990; Beyer et a!., 2015). For example, bocaccio 
(S. paucispinis ) range from the Southern California 
Bight to British Columbia, Canada; individuals in the 
Southern California Bight frequently exhibit multiple 
brooding and those in Central California occasionally 
exhibit multiple brooding, but the phenomenon has never 
been reported in individuals north of Cape Mendocino 
(He et aL, 2015b). Multiple brooding has not been 
reported in northern stocks (Love et aL, 2002; Conrath, 
2017), nor in species typically found in deeper water (on 
the continental slope at depths >400 m), other than the 
bank rockfish (<S. rufus ) in Southern California. However, 
beyond these general observations, the phenomenon of 
multiple brooding has never been empirically evaluated 
in relation to environmental or life history factors, and 
we have yet to determine if multiple brooding can be 
robustly predicted on a species level to allow estimation 
of the implications to stock assessments and climate vul¬ 
nerability assessments. 
We sought to quantitatively document the relationship 
between the phenomenon of multiple brooding in West 
Coast roekfishes with latitude or other demographic 
(maximum length, maximum age, and natural mortal¬ 
ity rate) and environmental (temperature and dissolved 
oxygen [DO] at depth, and average depth) factors and to 
determine which of these factors would be the best pre¬ 
dictors of multiple brooding. Specifically, we hypothesized 
that rockfish species of the continental shelf of the West 
Coast of the United States that are able to produce mul¬ 
tiple broods would be statistically more likely to inhabit 
lower latitudes (32-3S°N) or warmer v/aters (>9°C 
at depth). The results of this research are intended 
to provide a more comprehensive understanding of 
the reproductive biology of rockfish species and to 
improve the information available for their sustainable 
management. 
Materials and methods 
Study area and data collection 
Researchers at the Fisheries Ecology Division of the NOAA 
Southwest Fisheries Science Center have been collecting 
fecundity samples of a range of roekfishes to improve 
our understanding of size-dependent fecundity and care¬ 
fully noting the presence or absence of multiple broods 
(Beyer et aL, 2015; Lefebvre et aL, 2018). On the basis 
of these efforts and previous reports (Moser, 1966, 1967; 
MacGregor, 1970; Wyllie Echeverria, 1987; Love et aL, 
1990; Ralston and MacFarlane, 2010), we classified 
13 rockfish species found on the continental shelf of the 
West Coast as being multiple brooders and 11 shelf rock¬ 
fish species as single brooders (n-24; Table 1). 
The NOAA Northwest Fisheries Science Center has 
conducted a fishery-independent groundfish bottom- 
trawl survey along the U.S. West Coast annually since 
2003, in an area extending from the U.S.-Mexico bor¬ 
der (latitude 32°30 / N) to the U.S.-Canada border 
(latitude 48°10'N) and including depths from. 55 m to 
1280 m (Keller et aL, 2012). We downloaded physical 
(trawl location, temperature, DO concentration at depth, 
and trawl depth) and biological (species and sex) data 
for shelf roekfishes (as defined in PFMC, 2016) for trawl 
tows conducted from 2004 through 2015 (FRAM Data 
Warehouse, Northwest Fisheries Science Center, avail¬ 
able from website). It is important to note that the DO 
concentration at depth was not collected until 2007 and 
was not collected consistently until 2010. We downloaded 
commercial landings over the same time period from the 
Pacific Fisheries Information Network (PacFIN, avail¬ 
able from website) and assigned port code latitudes based 
on port locations. We acquired the maximum length (i.e., 
asymptotic average fork length in centimeters; see Fran¬ 
cis, 1988), maximum age (in years), and natural mortal¬ 
ity rate of the species of interest from published stock 
assessments by the Pacific Fishery Management Council 
(available from website), Dick and MacCall (2010), and 
Love et al. (2002) (Table 1). Estimates of the natural mor¬ 
tality rate for shortbelly rockfish (Sebastes jordani ) were 
used for halfbanded and pygmy roekfishes (S', semicinc- 
tus and S. wilsoni , respectively) because natural mortal¬ 
ity had not been estimated for those 2 species and all 3 
species are small, rapidly growing dwarf species. Species 
were excluded from our analyses if fewer than 20 individ¬ 
uals were collected over all the trawl samples; trawl sam¬ 
ples were excluded if the tow of the trawl was conducted 
deeper than the depth range of continental shelf habitat 
(depths >400 m). 
Data analysis 
To provide some context to our analyses, we explored 
how cumulative survey and commercial catch (measured 
as weight in kilograms) varied by latitude for each spe¬ 
cies by plotting the cumulative distribution of the sur¬ 
vey and commercial catch data, and the arithmetic mean 
