332 
Fishery Bulletin 117(4) 
that have experienced some level of historic fishing pres¬ 
sure, it is possible that the true maximum ages may be 
greater (Newman et al. 1 ; Newman and Dunk, 2003). The 
growth rates were also similar to those of other species 
of Pristipomoides, although the growth coefficients are 
slightly greater for both species. Estimates of and FL 
at age indicate that, although statistically significant, the 
differences between the sexes may be inconsequential 
because of the small differences in the estimates. 
Natural mortality, generally considered one of the most 
difficult population parameters to estimate, is also one 
of the most critical and influential parameters in stock 
assessments (Pauly, 1980; Brodziak et al., 2011b; Thorson 
et al., 2017). Direct, stock-specific methods, although 
preferred, are typically unfeasible because of prohibitive 
costs and data requirements. Because of these limita¬ 
tions, previous research on deepwater snappers (Newman 
et al. 1 , 2016; Newman and Dunk, 2003; Burton et al., 
2016; Williams et al., 2017) used indirect methods that 
use empirical and theoretical relationships (Beverton and 
Holt, 1959; Hoenig, 1983; Jensen, 1996; Then et al., 2015). 
These methods often suffer from low sample sizes and do 
not necessarily address spatial variability or taxonomic 
issues (Kenchington, 2014; Nadon and Ault, 2016). Esti¬ 
mating natural mortality by using a catch curve analysis 
with information from the unfished areas is considered 
an improvement over simply assigning a value of natural 
mortality (Jennings et al., 2001) or estimating it indirectly 
from empirical or theoretical relationships (Pauly, 1980; 
Hoenig, 1983; Then et al., 2015). In this study, however, 
the estimates of natural mortality obtained by using the 
multinomial catch curve analysis and by using the natural 
mortality estimator (maximum age method) are compara¬ 
ble, supporting the recommendation by Then et al. (2015) 
that their maximum age method provides the most reli¬ 
able indirect estimates of natural mortality. The results of 
our research also indicate that the maximum age method 
is applicable for estimating natural mortality of deepwa¬ 
ter snappers. 
Using age composition data from an unfished area pro¬ 
vides a direct estimate of natural mortality and allows the 
uncertainty in natural mortality to propagate to the total 
mortality estimate in a fished area (Maunder and Punt, 
2013). The quantification of uncertainty for natural mor¬ 
tality also means that it can serve as a prior distribution 
for natural mortality in future assessments. Simulations 
by Punt et al. (2001) and Punt and Methot (2004) demon¬ 
strated that data collection within unfished areas (e.g., 
marine protected areas [MPAs]) results in improvement in 
the estimation of natural mortality. Garrison et al. (2011), 
who simulated the value of MPAs to estimate life history 
parameters, reported that using data collected within 
MPAs was an improvement to estimating natural mortal¬ 
ity from fisheries data, although the success of estimating 
natural mortality was dependent on life history, move¬ 
ment, and model complexity. 
The results of our study highlights the value of obtain¬ 
ing age information for longer-lived species and not rely¬ 
ing solely on size information for parameter estimation or 
investigating fishing effects. In both the goldeneye jobfish 
and goldflag jobfish, the differences in length-based data 
(size composition and mean FL) between the fished and 
unfished areas were much smaller than the differences in 
age-based data (age composition and mean age). There¬ 
fore, inferences on fishery-related effects on populations 
may be underestimated or undetected if only length-based 
data are available. This is especially true for goldflag job¬ 
fish because the oldest fish (18 years) in the fished areas 
was sampled from Sarigan, the most remote island within 
the fished areas. The maximum age was 10 years around 
the populated island of Guam, which along with nearby 
seamounts and banks is the focal point of the fishery in 
the Mariana Archipelago. 
The poor correlation between length and age occurs for 
other deepwater snappers that reach at a young age 
relative to their maximum age (Newman et al., 2016). 
The results of this study indicate that age is a more sensi¬ 
tive and reliable indicator of deepwater snapper exploita¬ 
tion relative to size, likely because of decoupling of age and 
length. It is critical to consider this aspect when develop¬ 
ing monitoring programs, especially if both length and age 
information are available to assess stock status. 
This study’s findings also stress the importance of using 
appropriate life history values as inputs when using natu¬ 
ral mortality estimators. The differences in maximum age 
between the fished and unfished areas were large (golden¬ 
eye jobfish: 10 versus 28 years; goldflag jobfish: 18 versus 
32 years). Applying the Then et al. (2015) natural mortality 
estimator (maximum age method) to the fished areas led 
to a very different result, depending on which maximum 
age was used: estimates were close to 3 times greater for 
the fished area maximum age than for the unfished area 
maximum age. These different estimates of natural mor¬ 
tality contribute to very different exploitation ratios, which 
may result in inaccurate advice on stock status for fisher¬ 
ies managers. This potential for inaccuracy highlights the 
need for thoughtful application of natural mortality esti¬ 
mators and reliable estimates of maximum age when using 
these empirical methods. 
An important assumption of the spatial comparisons 
and the mortality estimates is that samples from the 
fished and unfished areas were from the same stock, with 
no movement of adults between areas. This assumption 
is valid because preliminary research indicates that, with 
very few exceptions, deepwater snappers generally do 
not move great distances (Weng, 2013; O’Malley 8 ). Also, 
growth of the deepwater snapper E. carbunculus var¬ 
ied across 20° of latitude in the Pacific Ocean (Williams 
et al., 2017). These authors suggest that these differences 
in growth were driven by the metabolic theory of ecology 
because of large differences in temperature. Although the 
latitudinal gradient across the Samoa Archipelago is very 
8 O’Malley, J. 2015. A review of the cooperative Hawaiian bottom- 
fish tagging program of the Pacific Islands Fisheries Science 
Center and the Pacific Islands Fisheries Group. NOAA, Natl. 
Mar. Fish. Serv., Pac. Isl. Fish. Sci. Cent. Admin. Rep. H-15-05, 
36 p. [Available from website.] 
