ICES Journal of Marine Science: Journal du Conseil

Farmer, Nicholas, A;Froeschke, John, T;Records, David, L

doi: 10.1093/icesjms/fsz238pmid: N/A

Abstract In a derby fishery, anglers race to catch as many fish as possible during a limited season. To meet legal mandates to prevent overfishing, forecasting accuracy is paramount. Red Snapper is among the most prized species in the US Gulf of Mexico and represents a politically charged derby fishery case study. We describe the management considerations, data, methods, and specialized statistical forecasting approaches used to estimate recreational component season lengths to maximize fishing opportunities while meeting mandates to constrain catch below legal limits. Retrospective analysis of model predictions for 2013–2017 indicated mean prediction error of 2626 ± 13 231, 3014 ± 15 744, and 42 975 ± 132 032 pounds whole weight per open day for charter, headboat, and private mode catch rates, respectively. Forecasting results using generalized linear models indicated that the annual harvest for 2017 would be caught in 2 d for the private angling component with an 18% probability of exceeding the component quota. The federal for-hire (charter and headboat) component season was estimated to be 48 d, with a 5% probability of exceeding the component quota. There is a broad scientific and management interest in identifying strategies to continue rebuilding the stock while increasing stakeholder access. Introduction In a regulated open-access commercial fishery for a high demand species, a total allowable catch, coupled with season controls, can result in a race to fish, as fishermen compete to harvest as many fish as possible before the season closes. Over time, this derby behaviour may lead to successively shorter seasons and result in problems such as depressed dockside prices and unsafe fishing practices (Conrad, 2010). Abbot et al. (2018) and Johnston et al. (2008) describe similar derby-style behaviour in recreational fisheries. As recreational seasons shrink, anglers may compress their fishing effort into shorter periods. This effort compression leads to higher daily catch rates, further reductions in season length, and negative impacts to angler welfare (Abbot et al., 2018). The compressed nature of a derby fishery requires specialized forecasting approaches, as biological, economic, environmental, and management covariates may impact catch rates or the average total catch obtained in a given day during the limited opening. Biological trends in the stock may influence the availability and size of fish, impacting both catch rates and mean weight of fish landed. Economic trends may influence anglers’ ability to fish, especially in federal waters, where fishing access may represent a significant expense due to fuel costs associated with the increased distance from shore. Environmental conditions may affect anglers’ ability to pursue particular species during limited season openings. Previously enacted management measures may affect angler perceptions of fishing opportunity; anglers may respond to shorter seasons by increasing effort during subsequent openings. Forecasting fish landings is a critical element in the management of fisheries stocks because it can inform strategy development and policy decisions on timelines necessary for effective management (Stergiou and Christou, 1996; Makridakis et al., 2008). Forecasts can be used to apply in-season or post-season accountability measures (AMs), including predicting closure dates. Forecasts can also provide a baseline for estimating the socio-economic impacts of proposed management actions. Previous work has suggested reliable forecasts of recreational catches are possible, especially when relatively consistent inter-annual and seasonal trends in catch rates were observed (Farmer and Froeschke, 2015). Here, we present a case study of the forecasting methods used to estimate Gulf Red Snapper federal recreational seasons. Because the factors contributing to the rise of a derby fishery are often due to management, our approach is necessarily case-specific. However, by forecasting recreational catch rates and season length within the context of changing regulations, increasing biomass, dynamic catch rates, and uncertain monitoring data, we hope to provide a suite of tools that will be useful to other scientists with similar forecasting needs. The objective of our study was to utilize historic information on state-specific catch rates for both the private angling and federal for-hire components along with covariates that impact recreational catch rates to predict catch rates for 2013–2017. Predicted federal catch rates for 2017 were used to predict the federal Red Snapper season length (i.e. number of days with federal waters open to retention of Red Snapper) for the recreational private and federal for-hire components, while accounting for predicted catches during proposed state seasons. A number of steps was needed to make forecasts: (i) compilation of fishery catch data, (ii) compilation and imputation of covariate data, (iii) phenomenological modelling of catch rate data by state using a broad suite of possible generalized linear model (GLM) regressions, and (iv) forecasting of days until the annual catch target (ACT) was reached. We evaluated a broad suite of possible GLM models to identify the best-fitting model with meaningful covariates for each state and component combination, evaluated the retrospective performance of the forecasting method, and applied our forecasts to predict the 2017 federal season. Methods Red Snapper is one of the most prized species in the Gulf for both commercial and recreational fishers. This species is managed by the Gulf of Mexico Fishery Management Council (Council) from the Florida Keys to south Texas within the federal waters of the US Exclusive Economic Zone (EEZ). In 2005, the Council established a new rebuilding plan for Red Snapper; dramatically reduced catch limits and the recreational bag limit, implemented a commercial individual fishing quota programme, and put controls on shrimp trawl fishing effort to reduce Red Snapper bycatch (GMFMC, 2006, 2007). Since the implementation of these additional management measures, Red Snapper overfishing has ended and the stock size has increased (Figure 1). Figure 1. Open in new tabDownload slide Changes in Red Snapper spawning potential, federal (black line) and state recreational season lengths, catch rates, and mean weights associated with the rebuilding of Red Snapper in the Eastern (Florida, Alabama, and Mississippi) and Western (Louisiana and Texas) Gulf of Mexico. Figure 1. Open in new tabDownload slide Changes in Red Snapper spawning potential, federal (black line) and state recreational season lengths, catch rates, and mean weights associated with the rebuilding of Red Snapper in the Eastern (Florida, Alabama, and Mississippi) and Western (Louisiana and Texas) Gulf of Mexico. The management structure for Red Snapper in 2017 included an overfishing limit (OFL), an acceptable biological catch (ABC), and two-component annual catch limits (ACLs) and ACTs (Figure 2). The OFL is defined as the maximum amount of a stock that can be caught in a year without resulting in overfishing. The ABC is a catch level defined as an amount reduced from the OFL by scientific uncertainty and risk preference (specified in a control rule) and recommended by the Council’s Scientific and Statistical Committee (SSC). A stock assessment incorporating total removals (landings + discards), indices of abundance, and age structure is used to determine population trajectories. The uncertainty from the assessment is used to inform the reduction from the OFL to the ABC. The ACL is the limit on the total annual catch for a stock or stock complex. The ACL is informed by the OFL and ABC and cannot exceed the ABC. The ACL can be set lower than the ABC to account for the degree to which the stock’s management measures are able to accurately constrain catch (i.e. management uncertainty) as well as relevant ecological, economic, and social considerations. The current Red Snapper total ACL is divided into commercial (51%) and recreational (49%) sector ACLs. The recreational sector ACL is sub-divided between two components: (i) private angling component (57.7%) and (ii) the charter vessel and headboat (federally permitted for-hire) component (42.3%). Figure 2. Open in new tabDownload slide Flow chart summarizing management and season length projection process for Gulf of Mexico red snapper. Figure 2. Open in new tabDownload slide Flow chart summarizing management and season length projection process for Gulf of Mexico red snapper. Based on the improving stock condition, the Red Snapper ABC has been increased incrementally from 5.0 million pounds whole weight (MP ww) in 2009 to 14.3 MP ww in 2015 (GMFMC, 2015). Managers must close the recreational sector when the ACL in state and federal waters combined is caught. Paradoxically, the improvement in stock condition has led to increased catch rates and mean weights, which have contributed to progressively shorter federal recreational fishing seasons (Figure 1). From 2004 to 2013, the recreational ACL, or quota, was exceeded seven times, despite a 78% reduction in season length (194 to 42 days) and a 20% increase in quota (4.47 to 5.39 MP; Figure 3). Due to these overages, from 2014 to 2018, an ACT was set for both components at 20% below the ACL, as a buffer to prevent the recreational sector from exceeding its ACL. The ACT is a management target, but there are no management consequences for exceeding it. If the ACL is exceeded, and the stock is in a rebuilding plan, then the ACL in the subsequent season will be reduced by the amount of the overage (i.e. a payback). Figure 3. Open in new tabDownload slide Red Snapper landings (millions of pounds, whole weight; MP ww) in MRFSS (1986–2012) and MRIP (2013–2016) units, relative to quota (dashed line). Figure 3. Open in new tabDownload slide Red Snapper landings (millions of pounds, whole weight; MP ww) in MRFSS (1986–2012) and MRIP (2013–2016) units, relative to quota (dashed line). The rebuilding of the stock coupled with the shortening of the season has resulted in a race to fish, further compounding the problem of effort compression (Figure 4). Pounds of fish landed per open federal day have increased sixfold since the implementation of the rebuilding plan in 2007. There have been no major changes in selectivity or federal size or bag limits; however, although the Council manages Red Snapper in federal waters, catches from state waters (i.e. from the shoreline to the federal waters boundary) count against the overall Gulf recreational quota for Red Snapper. The Gulf States set their own management measures for Red Snapper in state jurisdictional waters, including size limits, bag limits, and seasons. Historically, landings from state waters have represented about one-third of the red snapper total catch, with much lower catch rates in state waters per open day. In recent years, several Gulf states have extended their state waters seasons substantially beyond the federal season (Table 1). By 2016, the fishing season for the private recreational sector in federal waters was just 11 d; however, the state seasons for private recreational anglers (including for-hire vessels without federal permits) were much longer (Table 1). The increased catches from extended state seasons have effectively shortened the duration of the federal season (Figure 1: Season Length). Figure 4. Open in new tabDownload slide Pounds whole weight of Gulf of Mexico Red Snapper landed by the Red Snapper federal for-hire and private angling components relative to open federal days (open circles). Solid line denotes generalized linear model fit with Gamma distribution; dashed lines denote 95% confidence intervals. Note that the two components were unified as the “recreational sector” prior to 2015. Figure 4. Open in new tabDownload slide Pounds whole weight of Gulf of Mexico Red Snapper landed by the Red Snapper federal for-hire and private angling components relative to open federal days (open circles). Solid line denotes generalized linear model fit with Gamma distribution; dashed lines denote 95% confidence intervals. Note that the two components were unified as the “recreational sector” prior to 2015. Table 1. Approved and proposed (*) Gulf state water recreational Red Snapper regulations prior to the 2017 federal season. . State season (days) . Federal season (days) . . Year . FL . AL . MS . LA . TX . Federal . Private angling . Federal for-hire . Federal season dates . 2004 194 194 194 194 366 194 April 21 to Oct 31 2005 194 194 194 194 365 194 April 21 to Oct 31 2006 194 194 194 194 365 194 April 21 to Oct 31 2007 194 194 194 194 365 194 April 21 to Oct 31 2008 194 66 66 66 366 66 June 1 to Aug 5 2009 76 76 76 76 365 76 June 1 to Aug 15 2010 78 78 78 78 365 78 June 1 to July 24; Oct 1 to Nov 21 2011 48 48 48 48 365 48 June 1 to July 19 2012 46 46 46 46 366 46 June 1 to July 17 2013 58 42 42 113 365 42 June 1–28; Oct 1–14 2014 52 21 36 286 365 9 June 1–9 2015 70 41 118 215 365 10 44 June 1–10 (private angling component), June 1 to July 14 (federal for-hire component) 2016 85 66 102 272 366 11 46 June 1–11 (private angling component), June 1 to July 16 (federal for-hire component) 2017 78a 67 102* ≤334 365 TBD TBD . State season (days) . Federal season (days) . . Year . FL . AL . MS . LA . TX . Federal . Private angling . Federal for-hire . Federal season dates . 2004 194 194 194 194 366 194 April 21 to Oct 31 2005 194 194 194 194 365 194 April 21 to Oct 31 2006 194 194 194 194 365 194 April 21 to Oct 31 2007 194 194 194 194 365 194 April 21 to Oct 31 2008 194 66 66 66 366 66 June 1 to Aug 5 2009 76 76 76 76 365 76 June 1 to Aug 15 2010 78 78 78 78 365 78 June 1 to July 24; Oct 1 to Nov 21 2011 48 48 48 48 365 48 June 1 to July 19 2012 46 46 46 46 366 46 June 1 to July 17 2013 58 42 42 113 365 42 June 1–28; Oct 1–14 2014 52 21 36 286 365 9 June 1–9 2015 70 41 118 215 365 10 44 June 1–10 (private angling component), June 1 to July 14 (federal for-hire component) 2016 85 66 102 272 366 11 46 June 1–11 (private angling component), June 1 to July 16 (federal for-hire component) 2017 78a 67 102* ≤334 365 TBD TBD a Louisiana’s self-imposed recreational Red Snapper landing limit is based on 50% of the average percentages landed by each component between 1986 and 2013 (2010 excluded) and 50% of the average percentages landed by each component between 2006 and 2013 (2010 excluded). Under these computations, Louisiana’s historic share of the harvest for private anglers is 18.41% of the Gulf-wide harvest by private anglers and for charter 11.60% of the Gulf-wide harvest by the for-hire sector. In 2016, LA Creel estimated that Louisiana’s self-imposed recreational limit of 1 116 732 was exceeded by 1094 pounds. Therefore, for 2017, the self-imposed limit reflects a change in the Gulf-wide recreational ACL and removal of 2016 overages (715 171 + 330 290–1094 = 1 044 367 pounds). Open in new tab Table 1. Approved and proposed (*) Gulf state water recreational Red Snapper regulations prior to the 2017 federal season. . State season (days) . Federal season (days) . . Year . FL . AL . MS . LA . TX . Federal . Private angling . Federal for-hire . Federal season dates . 2004 194 194 194 194 366 194 April 21 to Oct 31 2005 194 194 194 194 365 194 April 21 to Oct 31 2006 194 194 194 194 365 194 April 21 to Oct 31 2007 194 194 194 194 365 194 April 21 to Oct 31 2008 194 66 66 66 366 66 June 1 to Aug 5 2009 76 76 76 76 365 76 June 1 to Aug 15 2010 78 78 78 78 365 78 June 1 to July 24; Oct 1 to Nov 21 2011 48 48 48 48 365 48 June 1 to July 19 2012 46 46 46 46 366 46 June 1 to July 17 2013 58 42 42 113 365 42 June 1–28; Oct 1–14 2014 52 21 36 286 365 9 June 1–9 2015 70 41 118 215 365 10 44 June 1–10 (private angling component), June 1 to July 14 (federal for-hire component) 2016 85 66 102 272 366 11 46 June 1–11 (private angling component), June 1 to July 16 (federal for-hire component) 2017 78a 67 102* ≤334 365 TBD TBD . State season (days) . Federal season (days) . . Year . FL . AL . MS . LA . TX . Federal . Private angling . Federal for-hire . Federal season dates . 2004 194 194 194 194 366 194 April 21 to Oct 31 2005 194 194 194 194 365 194 April 21 to Oct 31 2006 194 194 194 194 365 194 April 21 to Oct 31 2007 194 194 194 194 365 194 April 21 to Oct 31 2008 194 66 66 66 366 66 June 1 to Aug 5 2009 76 76 76 76 365 76 June 1 to Aug 15 2010 78 78 78 78 365 78 June 1 to July 24; Oct 1 to Nov 21 2011 48 48 48 48 365 48 June 1 to July 19 2012 46 46 46 46 366 46 June 1 to July 17 2013 58 42 42 113 365 42 June 1–28; Oct 1–14 2014 52 21 36 286 365 9 June 1–9 2015 70 41 118 215 365 10 44 June 1–10 (private angling component), June 1 to July 14 (federal for-hire component) 2016 85 66 102 272 366 11 46 June 1–11 (private angling component), June 1 to July 16 (federal for-hire component) 2017 78a 67 102* ≤334 365 TBD TBD a Louisiana’s self-imposed recreational Red Snapper landing limit is based on 50% of the average percentages landed by each component between 1986 and 2013 (2010 excluded) and 50% of the average percentages landed by each component between 2006 and 2013 (2010 excluded). Under these computations, Louisiana’s historic share of the harvest for private anglers is 18.41% of the Gulf-wide harvest by private anglers and for charter 11.60% of the Gulf-wide harvest by the for-hire sector. In 2016, LA Creel estimated that Louisiana’s self-imposed recreational limit of 1 116 732 was exceeded by 1094 pounds. Therefore, for 2017, the self-imposed limit reflects a change in the Gulf-wide recreational ACL and removal of 2016 overages (715 171 + 330 290–1094 = 1 044 367 pounds). Open in new tab Through 2017, federal regulations have opened the Red Snapper federal for-hire and private angler recreational fishing seasons each year on 1 June and closed the seasons when the respective recreational component ACTs were projected to be reached. These projections rely on ex-ante landings estimates, because current recreational fisheries catch reporting programmes are not timely enough for in-season monitoring. Although overfishing has ended and the stock is rebuilding, recreational fishermen are concerned about reduced access to Red Snapper in federal waters. The requirement to constrain the catch to the ACT, coupled with the goal of maximizing recreational fishing opportunities, necessitates an accurate forecast of the season length that achieves the full harvest of the ACT without exceeding the ACL. State regulations As federal seasons have shortened, the Gulf states have allowed progressively longer “state waters” fishing seasons. Federal waters extend from the state management jurisdiction boundary to the EEZ boundary. Projected state catches, before and after the federal season, are estimated in advance and deducted from the ACT, with the remainder available for the projected federal season. Although the Council and National Marine Fisheries Service encourage consistency with federal regulations, based on past precedent and pre-announced state seasons, this analysis assumed bag limits and size limits for Gulf states other than Texas would be consistent with the federal two-fish bag limit and 16-inch total length minimum size limit, but seasons for all Gulf states would be inconsistent with the federal season (Table 1). The projected private angling federal season lengths account for landings from state waters by private anglers and non-federally permitted charter vessels during state seasons to avoid a quota overage. Data sources Recreational Red Snapper federal and state season landings, catch rates, and mean weights from 2004 to 2017 were obtained from (i) the Marine Recreational Information Program (MRIP), (ii) the Southeast Region Headboat Survey (SRHS), (iii) the Louisiana Department of Wildlife and Fisheries creel survey (LA Creel), and (iv) the Texas Parks and Wildlife Department (TPWD). For‐hire and private angler Red Snapper landings from the MRIP survey are estimated using a combination of dockside intercepts (landings data) and phone surveys (effort data). Landings are estimated in both numbers and lb ww within 2‐month waves (e.g. Wave 1 = Jan/Feb … Wave 6 = Nov/Dec), area fished (state and federal waters), mode of fishing (charter, private angler mode/rental, shore), and state (west Florida, Alabama, Mississippi, and Louisiana). In 2013, MRIP implemented changes to the Access Point Angler Intercept Survey (APAIS). These changes to APAIS required a recalibration of historical landings to account for biases in the sampling time period; these re-calibrated landings were incorporated into the SEDAR-31 Update (2014) stock assessment and were used to generate the inputs for the season length projections in this report. Headboat landings are collected through logbooks completed by headboat operators and submitted to the SRHS. Landings (lb ww) are reported by vessel, day/month, and area. Landings from vessels participating in the 2014 and 2015 Headboat Collaborative Exempted Fishing Permit (HBC) were included in the SRHS data (http://sero.nmfs.noaa.gov/sustainable_fisheries/gulf_fisheries/reef_fish/2013/headboat_efp/). No estimates of uncertainty are generated by the SRHS; the magnitude of expansions between reported and expanded landings from the SRHS was used as a proxy for uncertainty. In recent years, these expansion factors were very small (<5%), as nearly all vessels were compliant with reporting requirements. Expanded headboat landings were obtained through 2017 from the Southeast Fisheries Science Center (SEFSC) Recreational ACL Dataset (accessed March 2019). LA Creel data were used in projections when MRIP estimates were not available from the ACL Data (i.e. private mode 2014 and 2016–2017, charter mode 2014–2017). Estimates of the number of Red Snapper landed in Louisiana (2014–2016) during the recreational season were provided by LA Creel. Estimated landings in weight were calculated by multiplying landings in numbers of fish for that season and sector by their average weight. Average weights were determined using biological samples from the month of the survey intercept week, or the entire year if samples from that month were not available. Variance in LA Creel mean weights was determined directly from their biological sampling data. The TPWD creel survey generates estimates of landings in numbers for charter vessels and private angler mode/rental boats fishing off Texas. Landings are reported in numbers by high- (15 May to 20 November) and low‐use (21 November to 14 May) time periods, area fished (state vs. federal waters), and mode of fishing (charter vs. private angler mode). To convert TPWD landings in numbers to landings in pounds, average lengths by mode, wave, and area fished were converted to weights (SEDAR-31 Update, 2014). TPWD staff provided preliminary 2016 landings as an input for the 2017 season projections, which were converted to MRIP wave equivalents by assigning mean weights to TPWD data using MRIP data from Louisiana, when available, or the other Gulf states (in aggregate) when MRIP data from Louisiana were not available. A similar approach was applied to TPWD mean weight for 2016 and for variance in TPWD weights for 2004–2016. Final TPWD landings for 2017 from the SEFSC Recreational ACL Dataset (accessed March 2019) were used for comparison with 2017 model predictions. For all sources, catch rates were expressed as landings in numbers of fish per open day by wave or week (if available). Uncertainty in MRIP, LA Creel, and TPWD mean estimates of average weights, numbers of fish landed, and pounds of fish landed are expressed as percent standard error and were converted to variance. When state seasons were inconsistent (e.g. different lengths, different start dates) with federal seasons, landings were partitioned into state and federal season landings. Landings were assumed to originate from state seasons if they were from waves during which the federal season was closed. For states other than Louisiana, if the federal season was open during a wave but a state season was open during days outside the federal season in that wave, federal season landings were considered to be federal waters landings plus a portion of the landings in state waters computed from the ratio of the federal season length in the wave to the state season length in the wave. For example, if the federal season were open for 30 d in a 60-d wave, and the state season was open for the entire wave, 100% of the landings originating from federal waters and 50% of the landings originating from state waters would be attributed to the federal season. If the state season ended before the federal season in a wave, then all landings were assumed to come from the federal season. For Louisiana, LA Creel weekly data were used to parse landings between in-season and out-of-season, assuming a uniform distribution of landings within the week. From 2015 to 2017, for-hire landings that occurred outside of the federal season were attributed to the state seasons and counted towards the private angling component’s quota, as they were assumed to originate from charter vessels without federal permits. All headboat landings were assumed to originate from the federal season, regardless of date landed. Projection modelling Projecting the red snapper federal component seasons required estimation of state private and charter landings and projection of federal landings that would count towards the federal private angling component ACT and projection of federally permitted headboat and charter landings that would count towards the federal for-hire component ACT (Figure 2). Forecasts of federal private angling and for-hire catch rates and average weights were developed for each state using generalized linear regression models fit to the 2004–2016 federal season data (Figure 2). Red Snapper harvest data for the regression models included wave 3 (May to June) private mode catch rates and waves 3–4 (May to Aug) charter and headboat catch rates from 2004 to 2016, excluding 2010 due to the confounding effects of harvest closures associated with the BP/Deepwater Horizon MC252 oil spill. The year 2014 was also excluded from the headboat data due to the confounding effects of the implementation of the HBC coupled with the short federal season. A suite of suitable GLMs for each state and mode were identified using an exhaustive search method from the glmulti library (Calcagno, 2013) in R v3.2.3 (R Core Team, 2014). As the goal of our study was predictive utility rather than identifying causal relationships or hypothesis testing, the best models for predicting the for-hire and private recreational seasons were identified by comparing prediction error for all models within two corrected Akaike Information Criterion points of the minimum (Akaike, 1973; Hurvich and Tsai, 1989) and using cross-validation with the cv.glm function in the boot package, which calculates the estimated K-fold cross-validation prediction error for generalized linear models (see Supplementary data: R Code; Canty and Ripley, 2016). Residual diagnostics were used to assess goodness-of-fit (see Supplementary data: Diagnostic Plots). Variable inflation factor tests were used to avoid multi-collinearity. The predictive covariates considered for regressions on federal season average weight and catch rates are presented in Figure 5 and described below: Figure 5. Open in new tabDownload slide Covariates considered in regression models for federal catch rates and average weights, including: (i) state season lengths, (ii) open weekend days, (iii) spawning stock biomass, (iv) fishable days for private angling component based on wind speeds <8.75 m/s, (v) fishable days for federal for-hire component based on wind speeds <8.75 m/s, (vi) Google Trends in searches for “Red Snapper”, (vii) fishable days for private angling component based on wave heights <1.2 m, (viii) fishable days for federal for-hire component based on wave heights <1.2 m, (ix) per capita gross domestic product (GDP), (x) mean Gulf of Mexico fuel price per gallon, (xi) year in the Red Snapper rebuilding plan, and (xii) Red Snapper recreational quota. Figure 5. Open in new tabDownload slide Covariates considered in regression models for federal catch rates and average weights, including: (i) state season lengths, (ii) open weekend days, (iii) spawning stock biomass, (iv) fishable days for private angling component based on wind speeds <8.75 m/s, (v) fishable days for federal for-hire component based on wind speeds <8.75 m/s, (vi) Google Trends in searches for “Red Snapper”, (vii) fishable days for private angling component based on wave heights <1.2 m, (viii) fishable days for federal for-hire component based on wave heights <1.2 m, (ix) per capita gross domestic product (GDP), (x) mean Gulf of Mexico fuel price per gallon, (xi) year in the Red Snapper rebuilding plan, and (xii) Red Snapper recreational quota. State Season Lengths—Based On State Announcements (2004–2016) and 2017 Announcements, or Statements Provided by State Fishery Managers (Table 1); Open Weekend Days—computed based on previous federal season lengths, with Fridays assigned a value of 0.5 and Saturdays and Sundays assigned a value of 1; the 2017 predicted value for regression fits was based on 2017 projected season lengths using 2016 catch rates and average weights only; Spawning Stock Biomass (SSB)—based on stock assessment model-estimated and model-projected SSB from SEDAR-31 Update (2014), included to account for changes in the size and abundance of Red Snapper, by region (i.e. East: Florida, Alabama, Mississippi; West: Louisiana, Texas) as the population rebuilds; Fuel Prices—because fuel prices may influence the willingness of recreational fishermen to fish farther offshore, mean June to July Gulf Coast (Petroleum Administration for Defense District 3) retail gasoline prices were obtained from U.S. Energy Information Administration (http://www.eia.gov/petroleum/data.cfm%23prices), with prices adjusted to 2010 US Dollars (USD) using the annual average Consumer Price Index (CPI) for all U.S. urban consumers (U.S. Department of Labor, 2017) and the 2017 June to July mean price predicted by extrapolating from the January to March 2017 value using the mean ratio for 2004–2016 of January to March to June to July prices; Year in the Rebuilding Plan—determined simply as the number of years since the rebuilding plan was revised in 2007, with all prior years coded as zeroes; Google Trends in Searches for “Red Snapper”—following Carter et al. (2015), based on Google Trends (2017) data from January to March with search counts adjusted by Google, dividing each point by the total searches of the geography and time range it represents, with relative popularity scaled on a range from 0 to 100; Per Capita Gross Domestic Product (GDP)—obtained, in 2010 adjusted USD, from historical and projected data from U.S. Department of Agriculture: Economic Research Service (2017a, b) because it is an indicator of the economic status of the United States overall, which may predict the ability of recreational anglers to afford to take trips for Red Snapper; Red Snapper Recreational Quota—obtained from NOAA from Federal Register notices; Year—the calendar year of fishing; Previous Year’s Average Weight or Catch Rate—the most recent value; Fishable Days Based on Weather—determined for each state using weather data from the NOAA National Data Buoy Center (U.S. Department of Commerce 2017a, b, c, d, e, f, g, h, i), with spatially explicit landings locations for Red Snapper from the Reef Fish Observer Programme and the NOAA Fisheries Bottom Longline Survey used to guide selection of data buoys with sufficient historical time series in areas near core federal Red Snapper catch locations off each Gulf state. Most buoys had data on hourly or half-hourly intervals from 2007 to 2016. For many states, multiple buoys were used (Supplementary Figure S1: Data Buoys). Beaufort Scale 5 measurements were used as a cut-off for weather suitable for offshore recreational fishing. Days with mean wind force measurements exceeding 8.75 m/s (17 knots) or mean wave heights exceeding 1.2 m (4 ft) were not considered fishable. The number of fishable days relative to the number of open days was determined for each fishing year. For private angling fishable days, only June measurements were used. For charter and headboat fishable days, June to July measurements were used. To impute missing values and generate predictions for fishable days in 2017, SARIMA models were fit to the data (Box et al., 2013). The SARIMA models were implemented using Proc ARIMA in SAS version 9.2 (SAS Institute) following approaches outlined in Farmer and Froeschke (2015). All possible combinations of single-difference SARIMA models for landings per fishable day by wave were considered (Supplementary Table S1: SARIMA Models). A single-difference SARIMA model only considers a maximum of one differencing term in the annual and one differencing term in the seasonal component. All SARIMA models were fit using conditional least squares. Stationarity tests were used to guide differencing selection. Final SARIMA model selection was guided by the examination of autocorrelations, inverse autocorrelations, partial autocorrelations, cross-correlations, residual diagnostics, and AIC. Goodness-of-fit for all SARIMA models exceeded R2 = 0.94. Parametric bootstrapping techniques were used to directly incorporate variance estimates from the surveys into the projection framework for all projections. The selected linear model function for each state and mode was iteratively fit to 1000 bootstrapped samples of input data for that state and mode based on the mean and variance for those observations. Bootstrapping treated annual catch and weight data as truncated normal distributions with a minimum of zero and a mean and standard deviation each sampling data source. Regression outputs included the mean and standard error for predicted mean weights and catch rates by state and mode. To avoid unrealistic estimates generated by overfitting models, any predicted catch rate greater than three times, or less than one-third of, the previous year’s catch rate was discarded, and the previous year’s catch rate, with uncertainty, was substituted as the forecast. Projected catch rates and mean weights To evaluate the performance of the projection methods described above, differences between projected and observed federal season catch rates were evaluated by year, state, and component, for 2013–2017. This retrospective analysis involved dropping the terminal year from the model, refitting the model following the methods described above, and comparing projection results to observed catch rates in the terminal year. Mean weights and catch rates in numbers of fish per day in the terminal year over 1000 bootstrap projections by year, state, and mode were forecast to the terminal year. The result of those forecasts (e.g. catch per day in pounds) was compared to observations from the terminal year that were similarly bootstrapped across 1000 bootstrapped runs to incorporate estimation error in observed catch rates. Prediction error was expressed as residuals (e.g. observed minus predicted catch rates) and as mean percentage error, computed as the sum of the residuals divided by the observations. Projected season length To estimate the length of the federal season to avoid an ACL overage, the following steps were taken: (i) estimate out-of-season state catches by mode, (ii) projection of federal in-season state catches by mode, and (iii) prediction of federal season length for 2017 (Figure 2). Federal season lengths for each component were estimated for 2017 based on anticipated state seasons (Table 1) and projected federal catch rates. Forecasts of catch rates and average weights from the best-fitting models were incorporated, along with their variance, into a Microsoft Excel-based season length projection model to determine the federal season length under each model scenario (Supplementary data: Season Projection Model). The season length projection model accounted for out-of-season catch rates. Catch rates and average weights for 2017 state seasons were estimated based on observed 2016 data, as 2016 was the most recent data available prior to the 2017 season and the only year where the state water jurisdiction for Red Snapper in Alabama, Mississippi, and Louisiana extended to 9 nmi (prior to the 2016 season, the U.S. Congress included language in the 2016 Department of Commerce Appropriations Act that extended the reef fish (including Red Snapper) management jurisdiction for Alabama, Mississippi, and Louisiana from 3 nmi from shore out to 9 nmi from shore, consistent with west Florida and Texas. This jurisdictional extension was made permanent by the 2017 Appropriations Act; Table 2). The model-projected state water catch rates and mean weights, with associated uncertainty, across 1000 bootstrapped runs. Bootstrapped state catch rates and average weights were estimated by wave and were multiplied by the days open per wave to estimate the total landings anticipated outside the federal season from Florida, Alabama, Mississippi, and Texas, along with the landings anticipated for Louisiana prior to the 1 June start of the federal season. Table 2. 2016 Gulf state water recreational Red Snapper average weights and catch rates used for 2017 federal season projections (mean ± SD). Wave . For-hire . Private . AL . FLW . LA . MS . TX . AL . FLW . LA . MS . TX . Mean Weight (lb ww)  Jan to Feb 8.9 ± 5.4 6.0 ± 3.6 9.0 ± 19.9 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 4.9 ± 5.2 6.0 ± 7.2 8.4 ± 24.5 7.3 ± 24.5  Mar to Apr 8.9 ± 10.3 6.0 ± 3.6 8.2 ± 15.3 6.4 ± 3.9 5.8 ± 10.3 6.5 ± 3.5 4.9 ± 5.2 7.0 ± 14.4 8.4 ± 11.6 4.9 ± 13.2  May to June 8.9 ± 5.4 5.8 ± 0.8 4.4 ± 5.7 5.7 ± 1.9 4.2 ± 2.1 8.3 ± 7.0 5.1 ± 6.5 7.1 ± 12.6 8.6 ± 3.8 4.7 ± 10.3  July to Aug 8.9 ± 6.0 5.9 ± 3.6 8.7 ± 9.7 6.4 ± 8.5 4.5 ± 10.4 6.5 ± 4.9 5.1 ± 3.1 8.6 ± 13.2 8.3 ± 8.6 4.3 ± 10.4  Sept to Oct 8.9 ± 5.7 6.6 ± 8.8 5.7 ± 13.0 6.4 ± 3.7 5.8 ± 11.1 6.5 ± 3.5 4.6 ± 10.1 7.0 ± 15.9 8.3 ± 10.8 4.4 ± 14.8  Nov to Dec 8.9 ± 5.4 6.0 ± 3.6 8.6 ± 14.1 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 5.1 ± 2.5 6.9 ± 6.3 8.4 ± 11.6 5.1 ± 2.5 Catch Rate (N/day)  Jan to Feb 0.0 ± 0.0 0.0 ± 0.0 62.9 ± 152.1 0.0 ± 0.0 115.9 ± 196.8 0.0 ± 0.0 0.0 ± 0.0 8361.1 ± 4199.0 0.0 ± 0.0 1409.3 ± 644.7  Mar to Apr 0.0 ± 0.0 0.0 ± 0.0 1621.7 ± 694.9 0.0 ± 0.0 540.6 ± 425.0 0.0 ± 0.0 0.0 ± 0.0 9434.9 ± 3046.6 0.0 ± 0.0 2611.6 ± 864.8  May to June 382.4 ± 243.8 2859.3 ± 1204.8 4181.0 ± 1277.3 11.8 ± 16.8 442.4 ± 384.5 1237.3 ± 782.6 97072.1 ± 33297.6 33285.9 ± 6700.9 5711.8 ± 3322.5 1868.6 ± 703.7  July to Aug 329.3 ± 226.2 5371.9 ± 4038.7 4605.7 ± 1028.9 21.4 ± 25.4 1548.8 ± 719.4 44852.3 ± 13506.6 14354.4 ± 5971.8 28802.0 ± 6136.8 8850.9 ± 4165.3 8031.6 ± 1640.4  Sept to Oct 0.0 ± 0.0 1250.0 ± 959.8 1851.5 ± 694.8 269.2 ± 289.3 522.1 ± 392.7 0.0 ± 0.0 81393.7 ± 33556.6 5924.2 ± 894.8 468.4 ± 429.6 1599.7 ± 769.8  Nov to Dec 0.0 ± 0.0 0.0 ± 0.0 3389.2 ± 866.2 0.0 ± 0.0 492.3 ± 405.6 0.0 ± 0.0 0.0 ± 0.0 16257.7 ± 8704.7 0.0 ± 0.0 431.6 ± 414.1 Wave . For-hire . Private . AL . FLW . LA . MS . TX . AL . FLW . LA . MS . TX . Mean Weight (lb ww)  Jan to Feb 8.9 ± 5.4 6.0 ± 3.6 9.0 ± 19.9 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 4.9 ± 5.2 6.0 ± 7.2 8.4 ± 24.5 7.3 ± 24.5  Mar to Apr 8.9 ± 10.3 6.0 ± 3.6 8.2 ± 15.3 6.4 ± 3.9 5.8 ± 10.3 6.5 ± 3.5 4.9 ± 5.2 7.0 ± 14.4 8.4 ± 11.6 4.9 ± 13.2  May to June 8.9 ± 5.4 5.8 ± 0.8 4.4 ± 5.7 5.7 ± 1.9 4.2 ± 2.1 8.3 ± 7.0 5.1 ± 6.5 7.1 ± 12.6 8.6 ± 3.8 4.7 ± 10.3  July to Aug 8.9 ± 6.0 5.9 ± 3.6 8.7 ± 9.7 6.4 ± 8.5 4.5 ± 10.4 6.5 ± 4.9 5.1 ± 3.1 8.6 ± 13.2 8.3 ± 8.6 4.3 ± 10.4  Sept to Oct 8.9 ± 5.7 6.6 ± 8.8 5.7 ± 13.0 6.4 ± 3.7 5.8 ± 11.1 6.5 ± 3.5 4.6 ± 10.1 7.0 ± 15.9 8.3 ± 10.8 4.4 ± 14.8  Nov to Dec 8.9 ± 5.4 6.0 ± 3.6 8.6 ± 14.1 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 5.1 ± 2.5 6.9 ± 6.3 8.4 ± 11.6 5.1 ± 2.5 Catch Rate (N/day)  Jan to Feb 0.0 ± 0.0 0.0 ± 0.0 62.9 ± 152.1 0.0 ± 0.0 115.9 ± 196.8 0.0 ± 0.0 0.0 ± 0.0 8361.1 ± 4199.0 0.0 ± 0.0 1409.3 ± 644.7  Mar to Apr 0.0 ± 0.0 0.0 ± 0.0 1621.7 ± 694.9 0.0 ± 0.0 540.6 ± 425.0 0.0 ± 0.0 0.0 ± 0.0 9434.9 ± 3046.6 0.0 ± 0.0 2611.6 ± 864.8  May to June 382.4 ± 243.8 2859.3 ± 1204.8 4181.0 ± 1277.3 11.8 ± 16.8 442.4 ± 384.5 1237.3 ± 782.6 97072.1 ± 33297.6 33285.9 ± 6700.9 5711.8 ± 3322.5 1868.6 ± 703.7  July to Aug 329.3 ± 226.2 5371.9 ± 4038.7 4605.7 ± 1028.9 21.4 ± 25.4 1548.8 ± 719.4 44852.3 ± 13506.6 14354.4 ± 5971.8 28802.0 ± 6136.8 8850.9 ± 4165.3 8031.6 ± 1640.4  Sept to Oct 0.0 ± 0.0 1250.0 ± 959.8 1851.5 ± 694.8 269.2 ± 289.3 522.1 ± 392.7 0.0 ± 0.0 81393.7 ± 33556.6 5924.2 ± 894.8 468.4 ± 429.6 1599.7 ± 769.8  Nov to Dec 0.0 ± 0.0 0.0 ± 0.0 3389.2 ± 866.2 0.0 ± 0.0 492.3 ± 405.6 0.0 ± 0.0 0.0 ± 0.0 16257.7 ± 8704.7 0.0 ± 0.0 431.6 ± 414.1 Open in new tab Table 2. 2016 Gulf state water recreational Red Snapper average weights and catch rates used for 2017 federal season projections (mean ± SD). Wave . For-hire . Private . AL . FLW . LA . MS . TX . AL . FLW . LA . MS . TX . Mean Weight (lb ww)  Jan to Feb 8.9 ± 5.4 6.0 ± 3.6 9.0 ± 19.9 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 4.9 ± 5.2 6.0 ± 7.2 8.4 ± 24.5 7.3 ± 24.5  Mar to Apr 8.9 ± 10.3 6.0 ± 3.6 8.2 ± 15.3 6.4 ± 3.9 5.8 ± 10.3 6.5 ± 3.5 4.9 ± 5.2 7.0 ± 14.4 8.4 ± 11.6 4.9 ± 13.2  May to June 8.9 ± 5.4 5.8 ± 0.8 4.4 ± 5.7 5.7 ± 1.9 4.2 ± 2.1 8.3 ± 7.0 5.1 ± 6.5 7.1 ± 12.6 8.6 ± 3.8 4.7 ± 10.3  July to Aug 8.9 ± 6.0 5.9 ± 3.6 8.7 ± 9.7 6.4 ± 8.5 4.5 ± 10.4 6.5 ± 4.9 5.1 ± 3.1 8.6 ± 13.2 8.3 ± 8.6 4.3 ± 10.4  Sept to Oct 8.9 ± 5.7 6.6 ± 8.8 5.7 ± 13.0 6.4 ± 3.7 5.8 ± 11.1 6.5 ± 3.5 4.6 ± 10.1 7.0 ± 15.9 8.3 ± 10.8 4.4 ± 14.8  Nov to Dec 8.9 ± 5.4 6.0 ± 3.6 8.6 ± 14.1 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 5.1 ± 2.5 6.9 ± 6.3 8.4 ± 11.6 5.1 ± 2.5 Catch Rate (N/day)  Jan to Feb 0.0 ± 0.0 0.0 ± 0.0 62.9 ± 152.1 0.0 ± 0.0 115.9 ± 196.8 0.0 ± 0.0 0.0 ± 0.0 8361.1 ± 4199.0 0.0 ± 0.0 1409.3 ± 644.7  Mar to Apr 0.0 ± 0.0 0.0 ± 0.0 1621.7 ± 694.9 0.0 ± 0.0 540.6 ± 425.0 0.0 ± 0.0 0.0 ± 0.0 9434.9 ± 3046.6 0.0 ± 0.0 2611.6 ± 864.8  May to June 382.4 ± 243.8 2859.3 ± 1204.8 4181.0 ± 1277.3 11.8 ± 16.8 442.4 ± 384.5 1237.3 ± 782.6 97072.1 ± 33297.6 33285.9 ± 6700.9 5711.8 ± 3322.5 1868.6 ± 703.7  July to Aug 329.3 ± 226.2 5371.9 ± 4038.7 4605.7 ± 1028.9 21.4 ± 25.4 1548.8 ± 719.4 44852.3 ± 13506.6 14354.4 ± 5971.8 28802.0 ± 6136.8 8850.9 ± 4165.3 8031.6 ± 1640.4  Sept to Oct 0.0 ± 0.0 1250.0 ± 959.8 1851.5 ± 694.8 269.2 ± 289.3 522.1 ± 392.7 0.0 ± 0.0 81393.7 ± 33556.6 5924.2 ± 894.8 468.4 ± 429.6 1599.7 ± 769.8  Nov to Dec 0.0 ± 0.0 0.0 ± 0.0 3389.2 ± 866.2 0.0 ± 0.0 492.3 ± 405.6 0.0 ± 0.0 0.0 ± 0.0 16257.7 ± 8704.7 0.0 ± 0.0 431.6 ± 414.1 Wave . For-hire . Private . AL . FLW . LA . MS . TX . AL . FLW . LA . MS . TX . Mean Weight (lb ww)  Jan to Feb 8.9 ± 5.4 6.0 ± 3.6 9.0 ± 19.9 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 4.9 ± 5.2 6.0 ± 7.2 8.4 ± 24.5 7.3 ± 24.5  Mar to Apr 8.9 ± 10.3 6.0 ± 3.6 8.2 ± 15.3 6.4 ± 3.9 5.8 ± 10.3 6.5 ± 3.5 4.9 ± 5.2 7.0 ± 14.4 8.4 ± 11.6 4.9 ± 13.2  May to June 8.9 ± 5.4 5.8 ± 0.8 4.4 ± 5.7 5.7 ± 1.9 4.2 ± 2.1 8.3 ± 7.0 5.1 ± 6.5 7.1 ± 12.6 8.6 ± 3.8 4.7 ± 10.3  July to Aug 8.9 ± 6.0 5.9 ± 3.6 8.7 ± 9.7 6.4 ± 8.5 4.5 ± 10.4 6.5 ± 4.9 5.1 ± 3.1 8.6 ± 13.2 8.3 ± 8.6 4.3 ± 10.4  Sept to Oct 8.9 ± 5.7 6.6 ± 8.8 5.7 ± 13.0 6.4 ± 3.7 5.8 ± 11.1 6.5 ± 3.5 4.6 ± 10.1 7.0 ± 15.9 8.3 ± 10.8 4.4 ± 14.8  Nov to Dec 8.9 ± 5.4 6.0 ± 3.6 8.6 ± 14.1 6.4 ± 3.9 5.8 ± 7.5 6.5 ± 3.5 5.1 ± 2.5 6.9 ± 6.3 8.4 ± 11.6 5.1 ± 2.5 Catch Rate (N/day)  Jan to Feb 0.0 ± 0.0 0.0 ± 0.0 62.9 ± 152.1 0.0 ± 0.0 115.9 ± 196.8 0.0 ± 0.0 0.0 ± 0.0 8361.1 ± 4199.0 0.0 ± 0.0 1409.3 ± 644.7  Mar to Apr 0.0 ± 0.0 0.0 ± 0.0 1621.7 ± 694.9 0.0 ± 0.0 540.6 ± 425.0 0.0 ± 0.0 0.0 ± 0.0 9434.9 ± 3046.6 0.0 ± 0.0 2611.6 ± 864.8  May to June 382.4 ± 243.8 2859.3 ± 1204.8 4181.0 ± 1277.3 11.8 ± 16.8 442.4 ± 384.5 1237.3 ± 782.6 97072.1 ± 33297.6 33285.9 ± 6700.9 5711.8 ± 3322.5 1868.6 ± 703.7  July to Aug 329.3 ± 226.2 5371.9 ± 4038.7 4605.7 ± 1028.9 21.4 ± 25.4 1548.8 ± 719.4 44852.3 ± 13506.6 14354.4 ± 5971.8 28802.0 ± 6136.8 8850.9 ± 4165.3 8031.6 ± 1640.4  Sept to Oct 0.0 ± 0.0 1250.0 ± 959.8 1851.5 ± 694.8 269.2 ± 289.3 522.1 ± 392.7 0.0 ± 0.0 81393.7 ± 33556.6 5924.2 ± 894.8 468.4 ± 429.6 1599.7 ± 769.8  Nov to Dec 0.0 ± 0.0 0.0 ± 0.0 3389.2 ± 866.2 0.0 ± 0.0 492.3 ± 405.6 0.0 ± 0.0 0.0 ± 0.0 16257.7 ± 8704.7 0.0 ± 0.0 431.6 ± 414.1 Open in new tab Federal projections were based on 1000 bootstrapped runs for each state and component under a user-specified federal season length for each component. Federal component landings were summed for each bootstrapped run across all states and added to projected state landings, including post-federal season landings anticipated for Louisiana to achieve their self-imposed combined (state and federal landings) quota (18.41% of private angler component ACL and 11.60% of federal for-hire component ACL). The mean component landings totals were compared to the component ACTs at different ACT buffers (i.e. 5%, 10%, 15%, and 20%), with the largest possible integer value of season length selected for each component that constrained mean catches below the ACT, while achieving within 1000 pounds of the Louisiana self-imposed quota over the course of their combined state and federal season. The risk of exceeding the ACL was evaluated at different ACT buffers by tabulating the percentage of bootstrapped runs that exceeded the ACL. Results Projected catch rates and mean weights AMs require that a component that exceeded its ACL in a year must have the component ACL and ACT in the next year reduced by the amount of the overage of the total ACL. Therefore, the private angling ACL and ACT for 2017 were adjusted for a 129 906 lb ww payback due to the ACL overage in 2016. The ACTs for 2017 were set at 2 278 000 and 3 004 075 lb ww for the federal for-hire and private angling components, respectively. Under the projected 2017 state seasons in Table 1 and the state average weights and catch rates from 2016 shown in Table 2, landings of Red Snapper from state seasons were forecast across the Gulf of Mexico as 2.42 ± 0.77 MP ww. Forecasts predicted 46 ± 27%, 11 ± 6%, 4 ± 3%, 26 ± 14%, and 6 ± 3% of landings would originate from Florida, Alabama, Mississippi, Louisiana, and Texas, respectively. Total state season landings were projected to account for 81 ± 26% of the private angling component ACT in 2017, with an average of 562 698 lb ww remaining for the federal water private angling component season. Rankings of regression model fits for the federal season are provided in Supplementary Tables: Weight Model Fits, Catch Model Fits. Selected model statistics are provided in Supplementary Table: Selected Model Stats. Residual deviance explained by covariates included in the final 2017 model configurations is provided in Table 3. Spawning stock biomass was the most consistently useful predictor for average weight trends, included in 47% of models. Weekend days were selected for all Louisiana mean weight models. Previous year’s catch rate was the most consistently selected predictor of trends in catch rates (73% of models); however, the per cent deviance explained by this covariate tended to be low. Year and year of rebuilding were selected in several models, as were economic indicators such as fuel price and per capita GDP. Table 3. Percent deviance explained by significant covariates in final 2017 regression model configurations, by state/region and mode (C: charter, P: private, H: headboat). . Florida . Alabama . Mississippi . Louisiana . Texas . Mode . C . P . H . C . P . H . C . P . H . C . P . H . C . P . H . Average weight  Year 35  Year of rebuilding 66  Weekend days 4 87 72 79  State days 89 11  SSB 90 98 90 93 80 74 91  Fuel price 4  Per capita GDP 27 40  Fishable days (WSP)  Fishable days (WH) 1 14  Google trends 4  Previous mean weight Catch rate  Year 86 87  Year of rebuilding 77 38  Weekend days 88  State days <1  SSB 27  Fuel price 18 4  Per capita GDP <1 54 66  Fishable days (WSP)  Fishable days (WH) 4  Google trends 78  Previous catch rate 5 6 3 <1 <1 28 19 19 1 15 2 . Florida . Alabama . Mississippi . Louisiana . Texas . Mode . C . P . H . C . P . H . C . P . H . C . P . H . C . P . H . Average weight  Year 35  Year of rebuilding 66  Weekend days 4 87 72 79  State days 89 11  SSB 90 98 90 93 80 74 91  Fuel price 4  Per capita GDP 27 40  Fishable days (WSP)  Fishable days (WH) 1 14  Google trends 4  Previous mean weight Catch rate  Year 86 87  Year of rebuilding 77 38  Weekend days 88  State days <1  SSB 27  Fuel price 18 4  Per capita GDP <1 54 66  Fishable days (WSP)  Fishable days (WH) 4  Google trends 78  Previous catch rate 5 6 3 <1 <1 28 19 19 1 15 2 Open in new tab Table 3. Percent deviance explained by significant covariates in final 2017 regression model configurations, by state/region and mode (C: charter, P: private, H: headboat). . Florida . Alabama . Mississippi . Louisiana . Texas . Mode . C . P . H . C . P . H . C . P . H . C . P . H . C . P . H . Average weight  Year 35  Year of rebuilding 66  Weekend days 4 87 72 79  State days 89 11  SSB 90 98 90 93 80 74 91  Fuel price 4  Per capita GDP 27 40  Fishable days (WSP)  Fishable days (WH) 1 14  Google trends 4  Previous mean weight Catch rate  Year 86 87  Year of rebuilding 77 38  Weekend days 88  State days <1  SSB 27  Fuel price 18 4  Per capita GDP <1 54 66  Fishable days (WSP)  Fishable days (WH) 4  Google trends 78  Previous catch rate 5 6 3 <1 <1 28 19 19 1 15 2 . Florida . Alabama . Mississippi . Louisiana . Texas . Mode . C . P . H . C . P . H . C . P . H . C . P . H . C . P . H . Average weight  Year 35  Year of rebuilding 66  Weekend days 4 87 72 79  State days 89 11  SSB 90 98 90 93 80 74 91  Fuel price 4  Per capita GDP 27 40  Fishable days (WSP)  Fishable days (WH) 1 14  Google trends 4  Previous mean weight Catch rate  Year 86 87  Year of rebuilding 77 38  Weekend days 88  State days <1  SSB 27  Fuel price 18 4  Per capita GDP <1 54 66  Fishable days (WSP)  Fishable days (WH) 4  Google trends 78  Previous catch rate 5 6 3 <1 <1 28 19 19 1 15 2 Open in new tab Mean weights were forecast to be stabilizing or declining across the Gulf. Example forecasts for Alabama are presented in Figures 6–8. In recent years, Alabama charter and private catch rates have shown a sawtooth pattern, with increases predicted in 2017 (Figures 6 and 7). Alabama headboat catch rates were forecast to be higher than previously observed (Figure 8). There was little variability in headboat input data in recent years due to near-perfect reporting compliance. Florida charter and private catch rates were projected to be relatively stable; headboat catch rates have declined since 2011 (Supplementary Figures Forecast Plots). Limited data were available for model-fitting in Mississippi. Louisiana and Texas charter catch rates were projected to be stable, but increases in catch rate were projected for private mode in both states, with the Louisiana increase resulting in automatic rejection and substitution of the previous year’s catch in the forecast. Louisiana headboat catch rates were forecast as a long-term average. A slight increase was forecast for Texas headboat catch rates. Figure 6. Open in new tabDownload slide Forecast for Alabama charter vessel mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Figure 6. Open in new tabDownload slide Forecast for Alabama charter vessel mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Figure 7. Open in new tabDownload slide Forecast for Alabama private angler mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Figure 7. Open in new tabDownload slide Forecast for Alabama private angler mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Figure 8. Open in new tabDownload slide Forecast for Alabama headboat mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Figure 8. Open in new tabDownload slide Forecast for Alabama headboat mode average weight and catch per day, showing basic model fits and bootstrapped run histograms accounting for uncertainty in weight and catch estimates. Retrospective analysis Retrospective analysis of model predictions for 2013–2017 indicated good predictive utility of the forecasting method through time (Figures 9 and 10). Prediction error for private catch rates was higher than for headboat or charter (Figure 9). Headboat catch rates for all states were underestimated in 2014 during the Headboat Collaborative Programme (Figure 10). Percent prediction error tended to be higher for Western Gulf states than Eastern Gulf states (Figure 10). Gulf-wide, mean prediction error averaged 2626 ± 13 231, 3014 ± 15 744, and 42 975 ± 132 032 pounds whole weight per open day for charter, headboat, and private mode catch rates, respectively. The distribution of model predictions tended to be well within the uncertainty around the observed catch rates for the private angling component (Figure 11). Gulf-wide headboat catch rates were overestimated in recent years (Figure 11). Figure 9. Open in new tabDownload slide Boxplot comparing prediction error (pounds whole weight) of observed vs. model-projected federal season catch rates in pounds whole weight per open day, by year, state, and mode of fishing. Model forecasting conducted by dropping terminal year, forecasting mean weights and catch rates in numbers of fish per day, and comparing the product of these forecasts to observations incorporating error over 1000 bootstrapped runs. Figure 9. Open in new tabDownload slide Boxplot comparing prediction error (pounds whole weight) of observed vs. model-projected federal season catch rates in pounds whole weight per open day, by year, state, and mode of fishing. Model forecasting conducted by dropping terminal year, forecasting mean weights and catch rates in numbers of fish per day, and comparing the product of these forecasts to observations incorporating error over 1000 bootstrapped runs. Figure 10. Open in new tabDownload slide Boxplot comparing per cent prediction error of observed vs. model-projected federal season catch rates in pounds whole weight per open day, by year, state, and mode of fishing. Model forecasting conducted by dropping terminal year, forecasting mean weights and catch rates in numbers of fish per day, and comparing the product of these forecasts to observations incorporating error over 1000 bootstrapped runs. Figure 10. Open in new tabDownload slide Boxplot comparing per cent prediction error of observed vs. model-projected federal season catch rates in pounds whole weight per open day, by year, state, and mode of fishing. Model forecasting conducted by dropping terminal year, forecasting mean weights and catch rates in numbers of fish per day, and comparing the product of these forecasts to observations incorporating error over 1000 bootstrapped runs. Figure 11. Open in new tabDownload slide Observed and predicted federal season catch rates (10 000 pounds whole weight per open day), by year and mode of fishing. Observed catch rates incorporate reporting error over 1000 bootstrapped runs. Predicted catch rates incorporate observation error and model error over 1000 bootstrapped runs. Figure 11. Open in new tabDownload slide Observed and predicted federal season catch rates (10 000 pounds whole weight per open day), by year and mode of fishing. Observed catch rates incorporate reporting error over 1000 bootstrapped runs. Predicted catch rates incorporate observation error and model error over 1000 bootstrapped runs. 2017 Projected season lengths The highest federal for-hire landings were estimated from Alabama and Florida. The highest federal private angling component landings were estimated from Alabama. The highest state private angling component landings were estimated from Florida and Louisiana. Using projected state and federal landings (Figure 12), the private angling component federal recreational season would be 2 d and the federal for-hire recreational season would be 48 d. At these season lengths, there would be an 18% probability of exceeding the private angling component ACT and a 5% probability of exceeding the federal for-hire component ACT (Figure 13). Combined, this provides an 8% overall probability of exceeding the total recreational ACL. Probability of exceeding the ACL increases with reduced ACT buffers (Figure 13). Without an ACT buffer, the private angling federal season could double in length; however, the probability of exceeding the component quota would increase substantially (Figure 13). The federal for-hire projections are more stable due to a levelling off in catch rates and average weights for most states and no need to account for state season landings. As such, the probability of exceeding the component quota for the federal for-hire component is much lower than for the private angling component at the same ACT buffers (Figure 13). Figure 12. Open in new tabDownload slide 2017 forecast landings using catch rates and weights using the best-fit regression models for each state, by jurisdiction and component. Figure 12. Open in new tabDownload slide 2017 forecast landings using catch rates and weights using the best-fit regression models for each state, by jurisdiction and component. Figure 13. Open in new tabDownload slide Mean season lengths and percentage of 1000 bootstrapped runs of projected landings that exceeded component quota across 1000 bootstrapped runs at different ACT buffer levels. Models configured with federal season lengths set to have mean landings equal to ACT. Error bars denote one standard deviation. Figure 13. Open in new tabDownload slide Mean season lengths and percentage of 1000 bootstrapped runs of projected landings that exceeded component quota across 1000 bootstrapped runs at different ACT buffer levels. Models configured with federal season lengths set to have mean landings equal to ACT. Error bars denote one standard deviation. Discussion The primary tool available to federal managers for controlling in-season catch rates is reducing the season length; however, reducing the season length does not proportionally reduce catch (Powers and Anson, 2016; Figure 4). It has long been recognized that bag limits and seasonal closures are relatively ineffective measures to control recreational exploitation rates (Cox et al., 2002). Resolving the issues that contribute to derby fishing may require a paradigm shift in regulatory approaches; however, it may be possible to better forecast catch rates and effectively set season lengths for fisheries that exhibit derby behaviour. In previous years, the Gulf Red Snapper recreational quota has been exceeded for a variety of reasons, including challenges with predicting angler behaviour and catch rates, inconsistent state regulations, and increasing fish sizes associated with a rebuilding stock (Shin et al., 2005). However, federal projection assumptions have been refined over time to better account for changes in average weights and daily catch rates. These refinements have led to increasingly more accurate predictions as described in NMFS-SERO (2013a, 2014, 2015, 2016). In addition, the implementation of a 20% buffer between the ACL and ACT has accounted for management uncertainty inherent in a protracted fishing season where the majority of landings are estimated by surveys and landings data are delivered to managers after the season ends. Some covariates were consistently useful predictors in regression models of Red Snapper federal season average weights and catch weights (Supplementary Table Deviance Explained). Spawning stock biomass was frequently a useful predictor of average weights; this is not surprising, as it is an indirect index of the exploitable biomass in the population. Previous year’s catch was the most commonly selected predictor for regression models on catch rate, which is a reasonable expectation in a volatile derby fishery. Year and year of rebuilding were useful predictors for catch rates; this is expected, as they indirectly account for both the rebuilding of Red Snapper with associated increases in the number of available fish along with the compression of recreational fishing effort into increasingly shorter fishing seasons. Per capita GDP was a useful predictor for private catch rates, possibly indicating more anglers on the water during years with favourable economic conditions. Fuel price was also a useful predictor, which may indirectly predict economic willingness to fish or financial ability to fish offshore where catch rates are anticipated to be higher, especially following heavy fishing pressure in state waters during the state season. Forecasts suggested some stabilization in mean weights and catch rates across the Gulf of Mexico. This may be due to trends in Red Snapper recruitment; reductions in Eastern Gulf spawning biomass (see SSB trend in Figure 1); or possible saturation in effort compression, where the season is sufficiently shortened that nearly all available anglers are pursuing red snapper during each day of the open season. In addition, extended state seasons may have decelerated effort compression in the federal season or reduced catch rates due to serial depletion of nearshore fishing areas. By explicitly accounting for uncertainty both in the input data and in the forecasting result, our approach may allow for a more dynamic treatment of the ACT buffer that could increase opportunities for federal waters fishing while still providing reasonable certainty that the ACL will not be exceeded. The forecast methods and approach described in this article represent a substantial improvement over past efforts to accomplish this goal. Future refinements should attempt to more explicitly incorporate angler behaviour and consider a broader suite of management strategies in an attempt to maximize social welfare while ensuring biological sustainability (Johnston et al., 2010; Lee et al., 2017). Using projection approaches similar to those described above, federal for-hire landings forecasts and season length projections have constrained catches below the ACT. Applying the ACL/ACT Control Rule (GMFMC, 2011), using federal for-hire landings data from 2014 to 2017, the Gulf Council recently reset the component ACT buffer for the federal for-hire component from 20% to 9% below the federal for-hire component ACL. The accuracy of federal for-hire projection approaches allowed a greater harvest while continuing to constrain landings to the component ACL (GMFMC, 2019). As with any projection model, the approaches discussed are dependent upon assumptions that historical data are unbiased and that historical trends are representative of future dynamics. Figure 4 indicates that catch rates increase as the stock rebuilds and the season is shortened. These dynamics are implicitly incorporated into the generalized linear regression approaches described by this document. Estimates of uncertainty in projections based on a single year of data do not account for inter-annual trends associated with the rebuilding of the stock or other covariates. They also do not account for any unanticipated changes in landings survey estimation methodology. It is critical to maintain “common currency” when projecting for management purposes. Maintaining a common currency for Red Snapper catch rates is especially challenging due to changes in data collection methods. For example, in 2013, MRIP replaced the Marine Recreational Fisheries Statistics Survey (MRFSS) programme as the primary methodology for collecting and estimating recreational catches in the Gulf. Data were back-calibrated for consistency; however, these calculations initially failed to account for adjustments to the dockside intercept protocol, resulting in the largest quota overrun on record (Figure 3). Subsequently, MRIP was replaced in Louisiana by the LA Creel survey and similar efforts to replace or modify MRIP are certified or underway in Mississippi (e.g. Tails-N-Scales Survey), Alabama (e.g. SnapperCheck), and Florida (e.g. Gulf Reef Fish Survey). Our projections for the 2017 federal season were based upon APAIS-adjusted MRIP, LA Creel, TPWD, and SRHS estimates. These sources were consistent with the data used to generate the ACL by the SEDAR-31 Update (2014) stock assessment and with the survey methods used to monitor catch levels relative to quotas in 2017. Prior to the development of rigorous projection methods such as those described above, Red Snapper recreational catches frequently overran the ACL (Figure 3). For the 2017 season, the greatest source of uncertainty was state season catch rates, as the states were projected to catch the majority of the ACL. Accounting for catches from state seasons is critical for accurate projections of the federal season. The percentage of private angling catch originating from state waters has increased substantially for most states in recent years as states have expanded their season lengths in state waters (Figure 1). Uncertainty in 2016 out-of-season state water catch rates and extension of state seasons following the announcement of the 2016 federal season were the primary reason for the private angling component quota overage in 2016. In 2016, state catch rates were underestimated for four of the five Gulf states; both Alabama and Florida announced longer seasons following the federal season announcement. On 14 June 2017, following the completion of these analyses and the close of the federal season on 3 June 2017, the U.S. Department of Commerce and the five Gulf States agreed to align Federal and State private angler Red Snapper fishing seasons for the remainder of the summer 2017. The Department of Commerce re-opened the 2017 private angler federal recreational season for 39 weekend days and holidays. While this action provided increased fishing opportunities for private anglers in the Gulf of Mexico, the 2017 private angling component quota was exceeded by 76%. Further complicating Red Snapper management, in 2018, the MRIP programme transitioned from a random digit dialling household survey to a mail-based survey using fishing license information and a mailing address database as its fishing effort sampling frame. The introduction of this new fishing effort survey requires back-calibration of all previously collected data and introduces an additional source of uncertainty when using historical data to predict future catch rates. For the 2018 and 2019 seasons, exempted fishing permits (EFPs) were issued to allow limited state management of recreational Red Snapper. These EFPs authorized the five Gulf state agencies, with certain conditions, to allow Red Snapper landed by the private angling component (include private vessel and state-licensed charter vessel anglers) within certain time periods determined by the respective states. In April 2019, the Council voted to delegate management authority of the private angling component for recreational red snapper fishing to each state. As our projection methods are state- and mode-specific, they may continue to be of use under this new management regime, and the federal for-hire season continues to be estimated by NMFS using approaches similar to those described herein. As more data from state seasons become available, these forecasting approaches may become useful to state managers attempting to set their own season lengths. Improvements upon this approach may explicitly incorporate the behavioural response of anglers into landings forecasts (Lee et al., 2017). In commercial fisheries, the adoption of property rights in the form of catch shares often ameliorates some of the problems associated with the race to fish (Casey et al., 1995; Asche et al., 2008; Agar et al. 2014). Researchers have investigated similar rights-based strategies for recreational fisheries, such as a catch share programme for Gulf of Mexico (Gulf) headboats (Abbot and Willard, 2017) or harvest tags for Gulf recreational fisheries (Johnston et al., 2008); however, full-scale implementation of such strategies is very limited. Cox et al. (2002) suggest bag limits and seasonal closures are ineffective to protect populations from overfishing in an open-access fishery and suggest effort and total-harvest-limitation programmes should be considered. Johnston et al. (2007) suggest the adoption of harvest tags have the potential to improve control over total catches, increase economic benefits and provide better information for fishery management. Absent the implementation of such management strategies, fisheries managers must do their best to maximize fishing opportunities within the bounds of the current management structure. Setting the season length is a crucial component of this and relies on accurate estimates of catch rates; a non-trivial task in fisheries that exhibit derby behaviour. The future management strategy of the Red Snapper fishery is uncertain. However, so long as there is substantial public interest in catching a limited resource, the potential for a derby fishery and the need to accurately control harvest through season closures remains. Acknowledgements Thanks to A. Strelcheck, M. Larkin, M. Levy, S. Gerhart, J. McGovern, J. Stephen, and J. Pulver for the review of this manuscript. Disclaimer The scientific results and conclusions, as well as any views or opinions expressed herein, are those of the author(s) and do not necessarily reflect those of NOAA or the Department of Commerce. References Abbott J. K. , Lloyd-Smith P., Willard D., Adamowicz W. 2018 . Status-quo management of marine recreational fisheries undermines angler welfare . Proceedings of the National Academy of Sciences of the United States of America , 115 : 8948 – 8953 . Google Scholar Crossref Search ADS PubMed WorldCat Abbott J. K. , Willard D. 2017 . Rights-based management for recreational for-hire fisheries: evidence from a policy trial . Fisheries Research , 196 : 106 – 116 . Google Scholar Crossref Search ADS WorldCat Agar J. J. , Stephen J. A., Strelcheck A., Diagne A. 2014 . The Gulf of Mexico Red Snapper IFQ program: the first five years . Marine Resource Economics , 29 : 177 – 198 . Google Scholar Crossref Search ADS WorldCat Akaike H. 1973 . Information theory and an extension of the maximum likelihood principle. In 2nd International Symposium on Information Theory, Tsahkadsor, Armenia, USSR, September 2–8, 1971 , pp. 267 – 281 . Ed. by Petrov B. N., Csáki F.. Akadémiai Kiadó , Budapest . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Asche F. , Eggert H., Gudmundsson E., Hoff A., Pascoe S. 2008 . Fisher’s behavior with individual vessel quotas; over-capacity and potential rent: five case studies . Marine Policy , 32 : 920 – 927 . Google Scholar Crossref Search ADS WorldCat Calcagno V. 2013 . glmulti: Model selection and multimodel inference made easy. R package version 1.0.7. https://CRAN.R-project.org/package=glmulti (last accessed 22 March 2019). Canty A. , Ripley B. 2016 . boot: Bootstrap R (S-Plus) Functions. R package version 1.3-18. https://cran.r-project.org/web/packages/boot/boot.pdf (last accessed 22 March 2019). Carter D. W. , Crosson S., Liese C. 2015 . Nowcasting intraseasonal recreational fishing harvest with internet search volume . PLoS One , 10 : e0137752 . Google Scholar Crossref Search ADS PubMed WorldCat Casey K. E. , Dewees C. M., Turris B. R., Wilen J. E. 1995 . The effects of individual vessel quotas in the British Columbia halibut fishery . Marine Resource Economics , 10 : 211 – 230 . Google Scholar Crossref Search ADS WorldCat Conrad J. M. 2010 . Resource Economics, 2nd edn, pp. 90–95 . Cambridge University Press , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Cox S. P. , Beard T. D., Walters C. 2002 . Harvest control in open-access sport fisheries: hot rod or asleep at the reel? Bulletin of Marine Science , 70 : 749 – 761 . OpenURL Placeholder Text WorldCat Farmer N. A. , Froeschke J. T. 2015 . Forecasting for recreational fisheries management: what’s the catch? North American Journal of Fisheries Management , 35 : 720 – 735 . Google Scholar Crossref Search ADS WorldCat Google Trends. 2017 . Interest Over Time (January 2004 to March 2017) for Search Term: Red Snapper. http://www.google.com/trends/ (last accessed 22 March 2019). Gulf of Mexico Fishery Management Council (GMFMC). 2006 . Amendment 26 to the Reef Fish Fishery Management Plan to Establish a Red Snapper Individual Fishing Quota Program . Gulf of Mexico Fishery Management Council , Tampa, Florida . 298 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gulf of Mexico Fishery Management Council (GMFMC). 2007 . Amendment 27 to the Reef Fish Fishery Management Plan and Amendment 14 to the Shrimp Fishery Management Plan . Gulf of Mexico Fishery Management Council , Tampa, Florida . 480 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gulf of Mexico Fishery Management Council (GMFMC). 2011 . Final Generic Annual Catch Limits/Accountability Measures Amendment for the Gulf of Mexico Fishery Management Council’s Red Drum, Reef Fish, Shrimp, Coral and Coral Reefs Fishery Management Plans, Including Environmental Impact Statement, Regulatory Impact Review, Regulatory Flexibility Analysis, and Fishery Impact Statement . Gulf of Mexico Fishery Management Council . Tampa, Florida . 378 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gulf of Mexico Fishery Management Council (GMFMC). 2014a . Recreational Accountability Measures for Red Snapper: Final Draft Framework Action for the Reef Fish Resources of the Gulf of Mexico . Gulf of Mexico Fishery Management Council , Tampa, Florida . 106 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gulf of Mexico Fishery Management Council (GMFMC). 2014b . Recreational Red Snapctor Separation: Final Amendment 40 to the Fishery Management Plan for the Reef Fish Resources of the Gulf of Mexico . Gulf of Mexico Fishery Management Council , Tampa, Florida . 306 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gulf of Mexico Fishery Management Council (GMFMC). 2015 . Red Snapper Quotas for 2015-2017+: Framework Action to the Fishery Management Plan for the Reef Fish Resources of the Gulf of Mexico including Draft Environmental Assessment, Regulatory Impact Review, and Regulatory Flexibility Act Analysis. Gulf of Mexico Fishery Management Council , Tampa, Florida . 73 pp. Gulf of Mexico Fishery Management Council (GMFMC). 2019 . Modification to the Recreational for-Hire Red Snapper Annual Catch Target Buffer: Framework Action to the Fishery Management Plan for the Reef Fish Resources of the Gulf of Mexico Including Draft Environmental Assessment, Regulatory Impact Review, and Regulatory Flexibility Act Analysis . Gulf of Mexico Fishery Management Council , Tampa, Florida . 34 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hurvich C. M. , Tsai C.-L. 1989 . Regression and time series model selection in small samples . Biometrika , 76 : 297 – 307 . Google Scholar Crossref Search ADS WorldCat Johnston F. D. , Arlinghaus R., Dieckmann U. 2010 . Diversity and complexity of angler behaviour drive socially optimal input and output regulations in a bioeconomic recreational-fisheries model . Canadian Journal of Fisheries and Aquatic Sciences , 67 : 1507 – 1531 . Google Scholar Crossref Search ADS WorldCat Johnston R. J. , Holland D. S., Maharaj V., Campson T. W. 2008 . Evaluation of Fish Tags as an Attenuated Rights-Based Management Approach for Gulf of Mexico Recreational Fisheries . Connecticut Sea Grant, Publication CTSG-08-07 , Groton, Connecticut . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lee M. , Steinback S., Wallmo K. 2017 . Applying a bioeconomic model to recreational fisheries management: groundfish in the Northeast United States . Marine Resource Economics , 32 : 191 – 216 . Google Scholar Crossref Search ADS WorldCat Makridakis S. , Wheelwright S. C., Hyndman R. J., 2008 . Forecasting Methods and Applications . Wiley , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2009 . 2009 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2009-02, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida. National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2011 . 2011 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2011-03, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida. National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2012 . 2012 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2012-01, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida. National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2013a . Updated 2013 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2012-02 (Addendum), NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida, May 21, 2013; updated June 4, 2013. National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2013b . 2014 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2013-10, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida, December 10, 2013. http://sero.nmfs.noaa.gov/sustainable_fisheries/gulf_fisheries/red_snapper/documents/pdfs/2014/2014_rs_rec_season.pdf (last accessed 22 March 2019). National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2014 . 2014 Gulf of Mexico Red Snapper Recreational Season Length Estimates (Revised and Updated). SERO-LAPP-2014-04, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida, April 21, 2014. http://sero.nmfs.noaa.gov/sustainable_fisheries/lapp_dm/documents/pdfs/2014/red_snapper_2014_season_length.pdf (last accessed 22 March 2019). National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2015 . 2015 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2015-04, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida, April 20, 2015. http://sero.nmfs.noaa.gov/sustainable_fisheries/lapp_dm/documents/pdfs/2015/rs_2015_rec_quota_projection.pdf (last accessed 22 March 2019). National Marine Fisheries Service Southeast Regional Office (NMFS-SERO). 2016 . 2016 Gulf of Mexico Red Snapper Recreational Season Length Estimates. SERO-LAPP-2014-04, NOAA Fisheries Service, Southeast Regional Office, St. Petersburg, Florida, May 2, 2016. http://sero.nmfs.noaa.gov/sustainable_fisheries/lapp_dm/documents/pdfs/2016/sero-lapp-2016-04_gom_red_snapper_rec_season_2016_20160502_sero_final.pdf (last accessed 22 March 2019). National Marine Fisheries Service Office of Science and Technology (NMFS-ST1). 2012 . MRFSS/MRIP Calibration Workshop Ad-hoc Working Group Report. NOAA Fisheries Service, Office of Science and Technology, Silver Spring, Maryland, May 16, 2012. http://sero.nmfs.noaa.gov/sustainable_fisheries/s_atl/sg/2013/reg_am13/documents/pdfs/reg13_app_d.pdf (last accessed 22 March 2019). National Marine Fisheries Service Office of Science and Technology [NMFS-ST1]. 2014 . MRIP Calibration Workshop II. Ed. by Carmichael J., Van Voorhees D.. NOAA Fisheries Service, Office of Science and Technology , Silver Spring, Maryland . https://www.st.nmfs.noaa.gov/Assets/recreational/pdf/MRIPCalibrationWorkshopII_FinalReport.pdf (last accessed 22 March 2019). Powers S. P. , Anson K. 2016 . Estimating recreational effort in the Gulf of Mexico Red Snapper Fishery using boat ramp cameras: reduction in federal season length does not proportionally reduce catch . North American Journal of Fisheries Management , 36 : 1156 – 1166 . Google Scholar Crossref Search ADS WorldCat R Core Team. 2014 . R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing , Vienna . www.R-project.org, Version 3.0.2. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Shin Y. J. , Rochet M. J., Jennings S., Field J. G., Gislason H. 2005 . Using size-based indicators to evaluate the ecosystem effects of fishing . ICES Journal of Marine Science , 62 : 384 – 396 . Google Scholar Crossref Search ADS WorldCat Southeast Data Assessment and Review (SEDAR-31). 2013 . Gulf of Mexico Red Snapper: Update Stock Assessment Report. SEDAR, North Charleston, SC. http://sedarweb.org/docs/sar/SEDAR%2031%20SAR-%20Gulf%20Red%20Snapper_sizereduced.pdf (last accessed 22 March 2019). Southeast Data Assessment and Review (SEDAR-31 Update). 2014 . Gulf of Mexico Red Snapper: Update Stock Assessment Report. SEDAR, North Charleston, South Carolina. http://sedarweb.org/docs/suar/SEDARUpdateRedSnapper2014_FINAL_9.15.2015.pdf (last accessed 22 March 2019). Stergiou K. , Christou E. 1996 . Modelling and forecasting annual fisheries catches: comparison of regression, univariate and multivariate time series methods . Fisheries Research , 25 : 105 – 138 . Google Scholar Crossref Search ADS WorldCat U.S. Department of Agriculture: Economic Research Service. 2017a . Real Per Capita GDP (2010 dollars) Historical [Data file]. https://www.ers.usda.gov/data-products/international-macroeconomic-data-set/international-macroeconomic-data-set/#Baseline%20Data%20Files (last accessed 22 March 2019). U.S. Department of Agriculture: Economic Research Service. 2017b . Real Per Capita GDP (2010 dollars) Projections [Data file]. https://www.ers.usda.gov/data-products/international-macroeconomic-data-set/international-macroeconomic-data-set/#Baseline%20Data%20Files (last accessed 22 March 2019). U.S. Department of Commerce. 2017a . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42012: Historical standard meteorological data, 2009-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42012 (last accessed 22 March 2019). U.S. Department of Commerce. 2017b . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42019: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42019 (last accessed 22 March 2019). U.S. Department of Commerce. 2017c . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42020: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42020 (last accessed 22 March 2019). U.S. Department of Commerce. 2017d . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42039: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42039 (last accessed 22 March 2019). U.S. Department of Commerce. 2017e . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42040: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42040 (last accessed 22 March 2019). U.S. Department of Commerce. 2017f . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42047: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42047 (last accessed 22 March 2019). U.S. Department of Commerce. 2017g . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station 42099: Historical standard meteorological data, 2007-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=42099 (last accessed 22 March 2019). U.S. Department of Commerce. 2017h . National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station BURL1: Historical standard meteorological data, 2004-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=burl1 (last accessed 22 March 2019). U.S. Department of Commerce. 2017 i. National Oceanic and Atmospheric Administration’s National Data Buoy Center. Station LOPL1: Historical standard meteorological data, 2011-2016 [Data files]. http://www.ndbc.noaa.gov/station_history.php? station=lopl1 (last accessed 22 March 2019). U.S. Department of Labor. 2017 . Bureau of Labor Statistics. Consumer Price Index for All Urban Consumers [Series: CUUR0000SA0]. https://data.bls.gov/timeseries/cuur0000sa0? series_id=cwur0000sa0 (last accessed 22 March 2019). U.S. Energy Information Administration. 2017 . Monthly Gulf Coast (PADD 3) Gasoline and Diesel Retail Prices [Data file]. http://www.eia.gov/dnav/pet/pet_pri_gnd_dcus_r30_m.htm (last accessed 22 March 2019). Published by International Council for the Exploration of the Sea 2019. This work is written by US Government employees and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)