Trends in SSB, Recruitment and Exploitation Rate - The trends in spawning stock biomass, recruitment, and exploitation rates from the continuity model results were similar to those from the SEDAR 33 assessment (Figure 22). Differences in SSB were greatest early and late in the time-series (Figure 22a). The estimated unfished condition of the continuity model was marginally larger than the SEDAR 33 model and the percent reduction describing the initial conditions (i.e., the recruitment offset parameter) was less resulting in a larger estimate of SSB in 1963 (Figure 22a). The SSB estimates of the two models were most similar between 1985 and 2010. Between 2010 and 2012, the estimates of SSB from the continuity model were less than those from the SEDAR 33 model. This corresponded to higher estimated exploitation rates later in the time-series than what was estimated by the SEDAR 33 model (Figure 22b). The estimated recruitment deviations and age-0 recruits were relatively similar throughout the time series, but deviates between 2007 and 2009. During this time, the estimates of the age-0 recruits from the continuity model were less than those estimated by the SEDAR 33 model (Figure 22c, d). Given the selectivity pattern (Figure 22), the 2007-2009 recruits should have been first captured by the recreational fleets within 3-5 years. Although the composition data used by the SEDAR 33 and continuity models have evidence of a 2006-2007 cohort the additional years of data (2013-2015) in the continuity model did not support the large predicted recruitment between 2007 and 2009 predicted by the SEDAR 33 assessment model. The selectivity curves for the commercial vertical line and recreational headboat fleets were also less domed, suggesting lower cryptic biomass than what was estimated by the SEDAR 33 model (Figure 23). - Fits to Indices - The fits to the relative indices of abundance are shown in Figure 24. Overall, the fits to the indices are similar, although some important deviations are noted. The SEDAR 33 assessment model fit the increasing trend in the last few years of the headboat index (2006-2010) more closely than the continuity model (Figure 24c). The SEDAR 33 model also overestimated the index values and predicted a greater increasing trend in the last few years of charter survey, private, and SEAMAP video survey indices of abundance (Figure 24d, e, g). The continuity model fit to the SEAMAP-video index is similar to SEDAR 33, but the fits to the headboat, charter, and private relative indices of abundance differ from SEDAR 33. The continuity model fit to the headboat index underestimates the increase in the last few years. It also underestimates the increase in the charter and private indices between 2005 and 2010, but better estimates the decline in these indices in more recent years. - Red tide mortality - The highest fishing mortality was estimated to happen in 2005 for the SEDAR 33 and continuity models. This high level of mortality is associated with additional mortality due to red tide. Its effect was modeled in terms of a discard fishery and caused a substantial increase in catch in 2005. The estimated red tide mortality expressed as an exploitation rate was 0.397 and 0.39 for the SEDAR 33 and the continuity models, respectively (Table 4). This corresponded to removals of ~3.4 million and 3.2 million gag in 2005. The terms of reference for this update assessment indicate that the potential effects of red tide should be re-evaluated with consideration of past red tide events and those of 2014 and 2015. Sensitivity runs with red tide events in 2005 and 2014 and 2005 and 2015 were completed. The estimated mortality due to red tide was higher in 2005 when the model allowed for a red tide event in 2014 or 2015 in the model (Table 4). The continuous mortality estimate was ~0.99 in 2005 and 2014 and the exploitation rates were 0.493 y-1 and 0.563 y-1. This resulted in approximately 5.1 million and 4.2 million gag removals due to red tide in 2005 and 2014, respectively. The continuous mortality estimate was ~0.81 in 2005 and 2015 and the exploitation rates were 0.425 y-1 and 0.492 y-1. This resulted in approximately 6.7 million and 10.4 million gag removals due to red tide in 2005 and 2015, respectively. - Continuity Projections and Stock Status - The reference points agreed upon by the GMFMC SSC for the SEDAR 33 gag grouper assessment were Fmax and SSBFmax, where SSB was defined to include only female biomass. The same definition of SSB and the same reference points were used for this update assessment. Figure 29 shows the yield-per recruit curves used to determine Fmax for the continuity model and the retrospective models. The maximum fishing mortality threshold (MFMT) is defined as Fmax and the minimum stock size threshold (MSST) was defined as (1-M)SSB Fmax. Fcurrent was calculated as the geometric mean of the last three years of the assessment (i.e., 2013-2015). Projections were run to evaluate stock status and provide OFL advice. Projections were run assuming that selectivity, discarding, and retention were the same as the three most recent years (2013-2015). Forecast recruitments are derived from the model estimated Beverton-Holt stock-recruitment relationship, based on the recent time period (i.e., 1984-2015). The catch allocation among fleets used for the projections reflects the average distribution of fishing intensity among fleets. Comparisons of the current fishing mortality and MFMT indicate the stock is not experiencing overfishing and comparisons of the current SSB and MSST indicate the stock is not overfished (Table 7). The management advice from the retrospective models provided similar status determinations. Each model was marginally more optimistic with the removal of a single year of data and indicated that the gag grouper population was not experiencing overfishing and was not overfished (Table 7). The retrospective models also have higher OFL streams than the full continuity model. This result suggests that we have consistently overestimated SSB and the allowable catch in recent years and the stock status may have not been as healthy as we previously estimated. - CONCLUSIONS - The assessment outcomes of the continuity model were similar to the SEDAR 33 benchmark assessment. Concern was expressed by fishermen after SEDAR 33 that the rapid increase in SSB seemed unreasonable given their catch rates. The increase in SSB in the continuity model was not as rapid as the SEDAR 33 trajectory, but the model predicts more of an increase than would be expected given the declining trends in the relative indices of abundance that continue through 2015. This, in combination with jitter results that illustrate the trade-offs in the data and the retrospective pattern indicate that the utility of management advice resulting from this model should be carefully evaluated. The alternative model was less optimistic than the continuity model. All things considered, this was a minor modification to the model, but the impact was significant with respect to stock status. The improvement in fit to the fishery-independent indices and the temporally comprehensive fishery dependent indices (i.e., charter and private) and the degradation in the fit to the commercial vertical line, the commercial longline, and headboat indices shows the sensitivity to the model and ultimately stock status to conflicting data sources. The jitter analysis indicated that the alternative model was more stable than the continuity model; however, the retrospective pattern was still prevalent. The persistence of the retrospective pattern may indicate that the model was over-parameterized (a total of 356 parameters were estimated) or that a time-variant mechanism was misspecified or ignored. These possibilities should be thoroughly examined during the next benchmark assessment of gag grouper.