Elsevier

Fisheries Research

Volume 244, December 2021, 106123
Fisheries Research

Shortfin mako hot sets – Defining high bycatch conditions as a basis for bycatch mitigation

https://doi.org/10.1016/j.fishres.2021.106123Get rights and content

Highlights

  • High shortfin mako bycatch in a longline fishery are important, impactful events.

  • Environmental conditions can be used to design mako bycatch mitigation strategies.

  • Presence/absence models of mako bycatch can be used to avoid potential high bycatch.

  • More complex models did not improve the ability to avoid mako bycatch.

Abstract

Shortfin mako sharks, Isurus oxyrinchus, were listed under Appendix II of CITES in 2019 in part due to the results of the last stock assessment for the North Atlantic population, which determined the population is overfished and experiencing overfishing. With population numbers expected to continue to decline, the managing body, the International Commission for the Conservation of Atlantic Tunas (ICCAT), has called for efforts to reduce shortfin mako bycatch. We evaluate the potential for reducing mako bycatch by identifying mako hot sets, those with particularly high shortfin mako bycatch. Environmental conditions were evaluated for their influence on catch per unit effort (CPUE) of shortfin mako sharks. Standardized CPUEs were calculated from the US pelagic longline observer program (2004–2012) using a generalized additive model (GAM) with a delta-lognormal approach applied to the environmental variables sea surface height, sea surface temperature, bathymetry, and chlorophyll-a concentration. Quantile regression (QR) was also performed to evaluate whether environmental variables can predict fishing conditions with high CPUE. The results of the GAM and QR methods were compared and assessed for their ability to predict and identify locations where shortfin mako CPUE is particularly high. The results suggest that using the binomial portion of the delta-lognormal model, the probability of positive bycatch, is the best basis to define an algorithm to avoid setting in conditions that might have high mako bycatch. Bycatch avoidance strategies built from probability of positive bycatch perform well enough at identifying hot sets to avoid half the shortfin mako bycatch with only a 20% reduction in effort.

Introduction

Shortfin mako sharks are highly migratory, long-lived sharks that are not known to aggregate. Shortfin makos tagged off the USA are distributed across the continental shelf and pelagic habitats following seasonal movements north and south that are correlated with seasonal patterns in ocean productivity (Queiroz et al., 2016, Vaudo et al., 2017). They prefer water temperatures 17–22 °C and appear to move inshore when the continental shelf waters warm, starting in April and May and move further east into the pelagic habitat in the fall and winter (Queiroz et al., 2016). Satellite tags also show that individual sharks exhibit high variability in their movements with some making long-distance migrations south into oligotrophic waters (Vaudo et al., 2017). Shortfin mako sharks’ distribution has high overlap with longline fleets (Queiroz et al., 2016, Queiroz et al., 2019).

In the Atlantic, shortfin makos are caught as bycatch in commercial longline fisheries targeting tunas and swordfish. They are assessed and managed by the International Commission for the Conservation of Atlantic Tunas (ICCAT) and historically have been one of the most commercially valuable sharks (Levesque, 2008). The population dynamics models used in the most recent stock assessment (Anonymous, 2017) and stock assessment update (Anonymous, 2019a) agree that the North Atlantic stock is overfished and experiencing overfishing. The ICCAT shark working group determined that the total shortfin mako catch needs to be reduced to less than 500 t from the level of 3115 t in 2017 to eliminate overfishing (Anonymous, 2019a). These findings contributed to the 2019 Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) listing the shortfin mako under Appendix II (Anonymous, 2019b). This listing will require that fishing nations demonstrate that fishing the shortfin mako would not threaten their chances for survival to obtain a CITES export permit for mako shark products (Anonymous, 1973). ICCAT recommendations and a listing under CITES acknowledge that shortfin mako sharks are not productive enough to rebuild without intervention.

This work looks specifically at the US portion of the pelagic longline fishery where shortfin mako are caught as bycatch but are not allowed to be retained. The US Pelagic Observer Program started in 1992 and monitors the US pelagic longline fleet operating from Newfoundland to Brazil, including the Caribbean and Gulf of Mexico (Beerkircher et al., 2004). The program is managed by NOAA’s Southeast Fisheries Science Center and has covered about 5% (1992–2001) to 8% (2002-present) of the vessels operating in this fishery. Observers on selected vessels record information about gear configurations and the species, size, sex, dead/alive status, and the ultimate fate (kept or discarded) of fish caught (Beerkircher et al., 2004). Bycatches of shortfin mako sharks recorded by observers in the US pelagic longline fishery are typically zero in most sets, so the shortfin mako bycatch per set data may be zero-inflated. When there is a positive shortfin mako bycatch, most of the time only 1 or 2 shortfin makos are caught per set. However, there are several events in the shortfin mako bycatch history that have caught upwards of 20 shortfin makos in one set. If we can identify the set of conditions that lead to sets with unusually high shortfin mako bycatch, hereafter referred to as “hot sets”, it may be possible to use this information to avoid high shortfin mako bycatch events and ultimately reduce the overall fishing mortality.

A generalized linear modeling (GLM) and generalized additive modeling (GAM) approach was used to predict mean shortfin mako bycatch per unit effort (CPUE), similar to the standardization methods used to generate abundance indices for past stock assessments (Cortes, 2007, Thorson et al., 2015). This study also determined conditions favoring particularly high CPUE, where much of the fishing mortality occurs. Quantile regression was used to focus on the upper tail of the CPUE distribution, rather than the mean, to reveal which conditions predict higher CPUE. Quantile regression is a method appropriate for abundance data with non-linear, non-symmetric, heterogenous scatter in response to a gradient (Koenker and Bassett, 1978, Cade and Noon, 2003, Anderson, 2008, Fukunaga et al., 2016). It addresses the scatter by allowing the coefficients to differ for the different parts of the abundance distribution (Koenker and Bassett, 1978) and has been used in ecology particularly to test whether environmental predictor variables predict areas of high or low abundance (Cade and Noon, 2003, Anderson, 2008, Fornaroli et al., 2016, Fukunaga et al., 2016). Application of quantile regression to the upper extreme allows the independent environmental variables to be viewed more easily as boundaries to suitable habitat (Fornaroli et al., 2016). Development of both model types involved a model selection process that included exploring categorical and numerical methods as well as GAMs with smoothers on the environmental variables to improve model fit and performance.

The traditional delta-lognormal approach (predicted probability of positive shortfin mako bycatch multiplied by the predicted mean of the positive shortfin mako bycatch to get the overall predicted mean CPUE) is hypothesized to perform poorly compared to the quantile regression fit to the upper extreme to predict hot sets. Quantile regression eliminates the need for a delta-lognormal approach and the difficulties in using zero-inflated data because the higher quantiles do not include zeros. For the purposes of determining conditions that favor a hot set, as opposed to standardized shortfin mako bycatch rates over time, quantile regression may be a better method. There is also more flexibility in model fit because there is no assumption of a normal distribution and the error distribution does not need to be specified. This method may provide the tools necessary to best identify potential hot sets and design a bycatch mitigation strategy.

This paper will present several ways to identify environmental conditions, regions and fishing methods that favor shortfin mako bycatch based on the outputs of the delta-lognormal model and quantile regression of the upper quantiles. The analysis will then determine which definition of a hot set and model could be used as a dynamic “no fish” algorithm to avoid shortfin mako bycatch and lower fishing mortality. The ultimate objective is to determine whether identifying conditions that favor high shortfin mako bycatch has the potential to decrease shortfin mako fishing mortality substantially while maintaining catches of target species.

Section snippets

Data preparation

Catch and effort data were obtained from the US pelagic longline observer program (1992–2016, 18,893 sets) (Southeast Fisheries Science Center, 2016), while weekly sea surface temperature composites on a 4 km grid (2003–2016) (GRIDDAP, 2018), daily sea surface height on a 0.25 degree grid (1992–2012) (Ducet et al., 2000), and daily chlorophyll-a concentrations (mg/m3) on a 4 km grid (2003–2021) (GRIDDAP, 2021) were downloaded from the NOAA CoastWatch satellite database and bathymetry on a 1

Results

In the US pelagic longline observer data, shortfin mako bycatch per set is generally low with a median of 0.0 and a mean of 0.68 ± 2.23 shortfin mako sharks. Most sets catch no shortfin mako sharks, or at most one or two (Fig. 1a). However, sets that catch one shortfin mako shark account for less than 18% of the total shortfin mako bycatch and sets that catch two or fewer shortfin mako account for less than 34% of the total shortfin mako bycatch (Fig. 1b). Sets with shortfin mako bycatches

Discussion

We presented several ways to identify environmental conditions, regions and fishing methods that favor high shortfin mako bycatch based on the outputs of the delta-lognormal model and quantile regression of the upper quantiles. We found that using the binomial portion of the delta-lognormal model, the probability of positive bycatch, was the best way to define a hot set basis for a “no fish” algorithm. Three potential mitigation strategies were designed based on the identification of hot sets

Conclusion

The shortfin mako catch in the pelagic longline fishery is highly clumped and patchy, with many sets catching none or one shark while some sets catch over 50 individuals at once. If most sets are not contributing to the bycatch rates, a method that focuses only on sets contributing to high bycatch rates would in theory be better, despite the added complexity. However, this was not the case; quantile regression, a more complicated method which can focus on high catch rates, did not perform

CRediT authorship contribution statement

Halie O’Farrell: Conceptualization, Methodology, Software, Formal analysis, Data curation, Writing – original draft, Writing – review & editing, Visualization, Elizabeth Babcock: Conceptualization, Methodology, Software, Resources, Writing – review & editing, Supervision, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

Thank you to Dr. David Die, Dr. Enric Cortes, Dr. Donald Olson, and Dr. John McManus for your scientific guidance and writing assistance. Thank you to the NOAA Observer Program and all observers, past and present, for collecting all the catch data used in this study. Thank you to NOAA Coastwatch, AVISO, and NASA's Goddard Space Flight Center for providing all the satellite data.

This project was funded by the NOAA Educational Partnership Program through the Living Marine Resources Cooperative

References (57)

  • C.A. Stow et al.

    Skill assessment for coupled biological/physical models of marine systems

    J. Mar. Syst.

    (2009)
  • F.J. Abascal et al.

    Movements and environmental preferences of the shortfin mako, Isurus oxyrinchus, in the southeastern Pacific Ocean

    Mar. Biol.

    (2011)
  • Anonymous, 1973. Text of the Convention. International Commission for the Conservation of Atlantic Tunas....
  • Anonymous

    Report of the 2017 ICCAT shortfin mako assessment meeting

    Collect. Vol. Sci. Pap. ICCAT

    (2017)
  • Anonymous

    Report of the 2019 shortfin mako shark stock assessment update meeting

    Collect. Vol. Sci. Pap. ICCAT

    (2019)
  • Anonymous, 2019b. Summary record of the fourth plenary session, Eighteenth meeting of the conference of the parties...
  • Beerkircher, L.R., Lee, D.W., Brown, C.J., Abercrombie, D.L., 2004. SEFSC pelagic observer program data summary for...
  • B.A. Block et al.

    Tracking apex marine predator movements in a dynamic ocean

    Nature

    (2011)
  • B.S. Cade et al.

    A gentle introduction to quantile regression for ecologists

    Front. Ecol. Environ.

    (2003)
  • J.G. Casey et al.

    Tagging Studies on the Shortfin Mako Shark (Isurus oxyrinchus) in the Western North Atlantic

    Aust. J. Mar. Freshw. Res.

    (1992)
  • E. Cortes

    Standardized catch rates of mako sharks from the U.S. Pelagic Longline Logbook and Observer Programs using a generalized linear mixed model

    Collect. Vol. Sci. Pap. ICCAT

    (2013)
  • E. Cortes

    Chondrichthyan demographic modelling: an essay on its use, abuse and future

    Mar. Freshw. Res.

    (2007)
  • J. Cramer et al.

    Pelagic longline bycatch

    Collect. Vol. Sci. Pap. ICCAT

    (1999)
  • E. Cuevas et al.

    Identification of potential sea turtle bycatch hotspots using a spatially explicit approach in the Yucatan Peninsula, Mexico

    Chelonian Conserv. Biol.

    (2018)
  • M. Drexler et al.

    Generalized additive models used to predict species abundance in the Gulf of Mexico: an ecosystem modeling tool

    Plos One

    (2013)
  • N. Ducet et al.

    Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2

    J. Geophys. Res.

    (2000)
  • J. Elith et al.

    Novel methods improve prediction of species’ distributions from occurrence data

    Ecography

    (2006)
  • M. Fasiolo et al.

    Fast calibrated additive quantile regression

    J. Am. Stat. Assoc.

    (2020)
  • Cited by (0)

    View full text