90 
Fishery Bulletin 119(1) 
L..~Lmax 200 Lyyat~Lamax Pairs, negative curvilinear rela- 
tionships between parameters in the K~L,, pair, and a 
positive curvilinear relationship between parameters in 
the M~K pair. The apparent outlier to this trend is a less 
positive, almost linear relationship between parameters 
in the M~K pair for grunts that could be an effect of the 
small sample size (n=9) for this taxonomic group. In this 
situation, the K parameter contains limited information 
on M, but the relationship still defines the range of M 
values that is likely for grunts. A similar situation was 
found for surgeonfishes and goatfishes by Nadon and 
Ault (2016). 
The shark group had a large standard deviation in the 
L..~Lyyax relationship. This difference was likely due to 
both the distribution of sizes for sharks and the decision 
to create the relationships for a taxonomic group broader 
than family. Our decision to group multiple families of 
sharks together was necessary because only one family of 
sharks, Carcharhinidae, had enough published life history 
information to generate family-specific relationships. We 
therefore extended the shark group in our study to com- 
prise all families in the orders Carcharhiniformes and 
Lamniformes, which include most families of conservation 
concern, such as requiem (Carcharhinidae), hammerhead 
(Sphyrnidae), thresher (Alopiidae), mackerel (Lamnidae), 
and sand tiger (Odontaspididae) sharks. After comparing 
the relationships for only species of Carcharhinidae with 
the relationships for all species of Carcharhiniformes and 
Lamniformes, we determined that it was appropriate to 
use a model for all families in the orders Carcharhini- 
formes and Lamniformes, given the current lack of data 
for species not in Carcharhinidae. As more life history data 
become available, it would be appropriate to create more 
family-specific models for other shark families. Until then, 
we believe our approach provides a reasonable option to 
assess shark stocks in data-poor situations. 
The average absolute percent difference in estimates 
of the SPR and of life history parameters from use of the 
stepwise approach and from life history studies for the 4 
test species was only 19%. This level of difference is simi- 
lar to the 22% average difference found by Nadon and Ault 
(2016). The differences between estimates from stepwise 
simulation compared favorably with average differences 
of 59% and 44% for genus- and family-level estimates 
from use of the FishLife approach, respectively. The model 
proved most successful in estimating L.., L,,a,, and M, with 
an average absolute difference of only 8-11% between the 
stepwise estimates and those from life history studies. It is 
not entirely surprising that it was a struggle to estimate 
L,, and L,,,, with the FishLife approach given that, in 
this approach, a local L,,,,, estimate is not used within an 
Lynax~L.. relationship to restrict L,, and L,,,4 estimates, as is 
done in the stepwise approach. The distributions of length 
parameter values estimated with the FishLife approach 
therefore effectively represent the range of values observed 
within a genus or family, a range that can be enormous (as 
evident in Figure 6). Wide distributions of length param- 
eter values limit the utility of the FishLife approach to 
population assessment models, such as the length-based 
spawning potential ratio model (Hordyk et al., 2016), that 
can be parameterized with the ratios of M to K and of L,,, 
to L,,. However, it is important to note that even these mod- 
els will require a reasonable estimate of L., to scale the 
length data used to fit model parameters. 
In addition, note that the distribution of these ratios 
can also be output from the stepwise simulation. It was a 
struggle to use our model to predict K, for which there was 
an average percent difference of 47% between the stepwise 
simulation and FishLife model, including the 124% dif- 
ference for redbreasted wrasse (K values were estimated 
with higher precision and accuracy when the FishLife 
approach was used; the K parameter was the only one for 
which FishLife outperformed the stepwise approach in both 
metrics). These large discrepancies in estimation of the K 
parameter did not have a strong effect on SPR estimation 
error because estimates of the SPR from use of the stepwise 
approach typically were within 0.05 of the value obtained 
from the relevant life history study. 
The study described here has resulted in the extension 
of our capability to conduct stock assessments in data- 
poor situations to a greater number of species. Our step- 
wise approach is generally more precise and accurate than 
the FishLife approach, although more research comparing 
both approaches is certainly warranted. It is important to 
reiterate that these meta-analytical approaches are not 
meant to replace life history studies, which typically have 
well-designed spatial coverage, appropriate gear selec- 
tion, large sample sizes, and peer-reviewed biological and 
statistical methods. We acknowledge that the stepwise 
approach presented here may introduce a large amount 
of uncertainty to any assessment relying on them, in com- 
parison with a species- and location-specific life history 
study. The stepwise approach is meant to allow temporary 
assessments to be implemented. We highly recommend 
full integration of the uncertainty in estimates of life 
history parameters throughout any assessment. We also 
recommend that stock status and management metrics 
be viewed in a risk assessment context, in which greater 
uncertainty is directly related to more conservative man- 
agement advice. Under these guidelines, predictions from 
the use of this approach have successfully been used to 
manage stocks in U.S. waters of the Pacific Ocean, pass- 
ing independent expert reviews (Nadon and Ault, 2016; 
Nadon, 2019). 
Acknowledgments 
This study was part of the NOAA Ernest F. Hollings 
Undergraduate Scholarship Program, and we would 
like to thank the NOAA Office of Education and the 
NOAA Pacific Island Fisheries Science Center, Fisheries 
Research and Monitoring Division, for facilitating the 
research. A. Yau was integral in facilitating the visiting 
researcher. B. Taylor greatly improved the project with his 
expertise and professional advice. We thank J. O’Malley, 
B. Langseth, and 2 anonymous reviewers for their import- 
ant feedback and comments on the manuscript. 
