A Bayesian approach to recover the theoretical temperature-dependent hatch date distribution from biased samples: the case of the common dolphinfish (Coryphaena hippurus)
Vicenç Moltó, Andres Ospina-Alvarez, Mark Gatt, Miquel Palmer, Ignacio A. Catalán
AA Bayesian approach to recover the theoretical temperature-dependent hatch datedistribution from biased samples: the case of the common dolphinfish(
Coryphaena hippurus ) (cid:73) Vicenc¸ Molt´o a , Andres Ospina-Alvarez a, ∗ , Mark Gatt b , Miquel Palmer a , Ignacio A. Catal´an a a Mediterranean Institute for Advanced Studies IMEDEA (UIB-CSIC), C / Miquel Marques 21, CP 07190 Esporles, Balearic Islands, Spain. b Department of Fisheries and Aquaculture, Fort San Lucjan, Triq il-Qajjenza, Malta.
Abstract
Reproductive phenology, growth and mortality rates are key ecological parameters that determine population dy-namics and are therefore of vital importance to stock assessment models for fisheries management. In many fishspecies, the spawning phenology is sensitive to environmental factors that modulate or trigger the spawning event,which di ff er between regions and seasons. In addition, climate change may also alter patterns of reproductive phe-nology at the community level. Usually, hatch-date distributions are determined back-calculating the age estimatedon calcified structures from the capture date. However, these estimated distributions could be biased due to mortalityprocesses or time spaced samplings derived from fishery. Here, we present a Bayesian approach that functions as apredictive model for the hatching date of individuals from a fishery-dependent sampling with temporal biases. Weshow that the shape and shift of the observed distribution is corrected. This model can be applied in fisheries withmultiple cohorts, for species with a wide geographical distribution and living under contrasting environmental regimesand individuals with di ff erent life histories such as thermo-dependent growth, length-dependent mortality rates, etc.
1. Introduction
Reproductive phenology, growth and mortality ratesare key ecological parameters that determine populationdynamics and are therefore of vital importance to stockassessment models for fisheries management (Lee et al.,2011; Methot and Wetzel, 2013; Stawitz et al., 2019;Walker et al., 2019). These parameters are di ffi cult toestimate and are not known for all fisheries, or evenapplied in the stock assessment models, especially indata-poor fisheries. This fact could lead to considerablesources of bias for the stock assessment models.The spawning phenology is sensitive to environmen-tal factors that modulate or trigger the spawning event,which di ff er among regions and seasons (P¨ortner andPeck, 2010; Poloczanska et al., 2016), climate changecan also alter the patterns of reproductive phenology atthe community level (Pankhurst and Munday, 2011; An-derson et al., 2013). If optimal condition (food avail- (cid:73) c (cid:13) // creativecommons.org / licenses / by-nc-nd / / ∗ Corresponding author
Email address: [email protected] (Andres Ospina-Alvarez ) ability, lower depredation rates) are observed at the be-ginning of the spawning season, growth and conditionis early promoted and survival to juvenile stages is in-creased (Lapolla and Buckley, 2005; Islam et al., 2015).On the contrary, when optimal environmental factorsoccur later in the spawning season, mortality ratesare considerable (e.g., higher accumulated depredationpressure, stronger e ff ects of advective or dispersive pro-cesses). In this situation, being born later in the spawn-ing season could be advantageous (Yoklavich and Bai-ley, 1990; McGovern and Olney, 2011; Xie and Watan-abe, 2005; Folkvord et al., 2015). One way to deter-mine the hatching date distribution is back-calculatingthe age from hard structures of fish (e.g., otoliths cap-tured in a fishery or scientific survey). However, theinferred distribution of observed birth-dates distributioncould be incorrect due to sampling time bias. The earliera fish is born, the more likelihood there is that it will diebefore the sampling date (Campana and Jones, 1992).Fishery-dependent sampling typically su ff ers from sucha temporal bias in the observed hatch-day distributionbecause fish are vulnerable only after reaching a cer-tain size / age threshold or simply due to the existenceof a mandatory beginning of the fishing season. Ac- Preprint submitted to ArXiv June 23, 2020 a r X i v : . [ q - b i o . O T ] J un ordingly, the spawning probability can only be prop-erly estimated from the observed birthday distributionafter incorporating the e ff ect of mortality rates.We present here a Bayesian approach that functionsas a predictive model for the hatching date of a popula-tion that is sampled with a temporal bias. Our model forpredicting the hatching date can be applied in fisherieswith multiple cohorts, for species with a wide geograph-ical distribution and living under contrasting environ-mental regimes and individuals with di ff erent life his-tories (e.g. thermodependent growth, length-dependentmortality rates, etc.). Therefore, the work presentedhere is relevant to fisheries management, ecology andbiology. The model can be incorporated into habitat pre-diction models, individual-based growth models, bio-energetic models, models projecting the impact of cli-mate on fish growth, etc. In this example, we apply themodel to the dolphinfish ( Coryphaena hippurus ) fish-ery in the NW Mediterranean (CopeMed, II, 2016). Thefishery is based on age-0 individuals that are spawned insummer in the area and are fished between August andDecember (CopeMed, II, 2016) after an extremely fastgrowth (Massut´ı et al., 1999).
2. Methods
Three main data sources were required to assess theseasonal pattern of spawning probability of dolphinfish:DB1. A database of juveniles’ daily otolith readings;DB2. A database of gonadosomatic index (GSI) estima-tions (indicative of average population spawning state)around the world; and DB3. Sea Surface Temperature(SST) estimations extracted from satellite imagery foreach location and date sorted in DB2.The rationale behind coupling these three databasesrespond to: 1) The hatch date distribution of the sam-pled fish should be related with the hatch date distribu-tion of the cohort (Methot, 1983); 2) The seasonal pat-tern of GSI should be a proxy of the seasonal spawningpattern (McQuinn, 1989); 3) Fish maturity and GSI arestrongly related with the thermal characteristics of theirspawning habitat (Dobson and Dodd, 1977; Neuheimerand MacKenzie, 2014; Pankhurst and Munday, 2011).Therefore, two sub-models can be built: The first onepredicts the probability
Prob. birth i , j that an individual i belonging to the cohort j was be born on the date hatch i , j ; and the second one to predict the fraction ofthe population being at the spawning state Prob. GSI t , j at the month t . Both sub-models are related to the tem-poral location of the spawning peak ( µ j ) and the spreadof the distribution of the actual dates of birth ( σ j ). Box 1. Databases
DB1. Juveniles’ otolith readings database1074 individuals, with recorded capture date,corresponding to 13 Mediterranean cohorts (3years from Balearic Islands and 8 years fromMalta; Table A.1).DB2. Monthly population-averaged GSI databasePopulation-averaged GSI and date for 7 popula-tions along a wide latitudinal gradient (Taiwan,Malta , Balearic Islands, North Carolina, SouthWest India, Japan and California; Table B.2)DB3. Monthly averaged SSTSST from satellite imagery for the locations (aver-ages of 100x100 km around the areas of capture)and dates from DB2 (Not shown here). i , j sub-model The hatch-date (birth) distribution sub-model is de-rived with the assumption of the hatch dates of a givencohort are normally distributed in time around a cohort-specific ( j ) peak date ( µ j ; Julian days from January 1st),and a cohort-specific spread ( σ j ; days). The fish i hasbeen actually captured at a given date and has a knownage and the probability can be approximated by:Prob . birth i , j = dnorm (cid:16) hatch i , j , mean = ˆ µ j , sd = σ j (cid:17) pnorm(capture i , j − age-min i , j , mean = ˆ µ j , sd = σ j (cid:17) (1)where dnorm denotes the probability density functionand pnorm denotes the cumulative probability functionof the normal distribution; hatch i j is the observed (in-ferred from the otolith) hatch date of the i fish fromthe j cohort; capture i j is the capture date (Julian days); age-min i j is the age from which the fish i reaches athreshold size after which it is considered vulnerable tofishing (20 cm; which is the approximate average lengthof the minimum size at catch in the available fishery-dependent data series (DB1)); and therefore, when thefishing mortality acts. This age is estimated using agrowth model developed for this species (Molt´o et al.,submitted). Finally, ˆ µ j represents the apparent spawn-2ng peak after being distorted by natural and fishingmortality, which is given by:ˆ µ j = µ j + m σ j (2)where µ j is the true peak location (Julian days) of thecohort j and m is the instantaneous natural mortality rate(1 .
703 day − ), estimated from the HoenigNLS empiricalprediction model using the maximum known life span(Tmax) for a given species as it was the best empiricalestimator of natural mortality, according to a study ofmore than 200 species (Then et al., 2014). Here, Tmaxwas set as 3 for dolphinfish (Massut´ı et al., 1999).At the between-cohort level, the spawning peak loca-tion ( µ j ) and the spread ( σ j ) are assumed to be a linearfunction of the sinusoidal temperature profile of a givenyear and region: µ j = β o + β Temp. Phase j + ε j (3) σ j = γ o + γ Temp. Mean j (4)where (cid:15) j is either a normally distributed error withzero mean and standard deviations sd .µ (Eq. 3) or σ j is assumed to be gamma distributed with a rate sd.rate (Eq. 4). Temp.Mean j , Temp.Amp j and Temp.Phase j arethe mean, amplitude and phase of a sinusoidal functionfitted to the temperature profile that has experienced thecohort j :Temp t , j = Temp. Mean j + Temp. Amp j sin (cid:16) π t / + Temp. Phase j (cid:17) + ε j (5) i , j sub-model The spawning state (GSI) distribution sub-model isderived from the gonadosomatic index data. The ob-served
GSI t , j (average GSI of a sample of fish from thepopulation j at the month t , in Julian days) is modelledby: GSI t , j = GSI max j Prob. GSI t , j + ε j (6)where GSI max j is the maximum GSI value attain-able at the population j (scale factor); Prob. GSI t , j is thefraction of the population being at the spawning stateat the month t ; and (cid:15) j is a normally distributed errorwith zero mean and standard deviations sd. GSI . In turn, Prob. GSI t , j is related with the spawning peak location ( µ j ) and the spread ( σ j ) of the actual hatch dates distri-bution (common parameters shared with Prob. birth i , j sub-model) by a super Gaussian distribution modelDecker (1994), that only di ff ers with the normal dis-tribution by the exponent POW . When
POW <
1, theresulting distribution has a pointed peak and long tails(”Christmas tree distribution”).
Prob. GSI t , j = σ j √ π e − ( t − µ j ) σ j POW (7)Therefore, eq. 7 e ff ectively links the (unbiased) prob-ability of spawning with the observed temperature pro-file, and the relationship between GSI and temperature.The parameters of the two sub-models above have beenestimated using a Bayesian approach implemented ina custom R script (R Core Team, 2019) (Appendix 1)that runs JAGS (Plummer, 2003) for moving the MCMCchains. We use the ”ones trick” (Kruschke, 2015) forsampling from the non-standard distribution describedin Eq. 1. Three independent chains were run. Theconvergence of the MCMC chains was assessed by vi-sual inspection of the chains and was tested using theGelman-Rubin statistic (Plummer, 2008).A threshold value of 1.1 or less was assumed to sug-gest convergence (Gelman et al., 2013). Posterior dis-tribution was estimated by at least 3 000 valid iterationsafter appropriate burning and thinning (one out 10 iter-ation were kept). The first 10 000 iterations were notconsidered. Nearly uninformative priors have been as-sumed for all the parameters. Specifically, priors for β (Eq. 3) and γ (Eq. 4) parameters are assumed to be nor-mally distributed with zero mean and tolerance = e − ,GSI.scale are assumed to be normally distributed withzero mean and tolerance = e − , but constrained to bepositive, and all tolerances and rates ( sd. rate in Eq. 4)are assumed to be gamma distributed with parametersshape = .
01 and scale = . / zenodo.3725530
3. Results
The theoretical (corrected) hatch-date distributionhas been estimated for the Mediterranean dolphinfishcohorts above mentioned (DB1). As an example, thecorrected hatch-date distribution for the Balearic Islands2004 cohort is shifted to the left (early in the spawningseason) with respect to the observed hatch-date distri-bution, with a di ff erence of 5 days in their median val-3es (Fig. 1). Our model also shows an adequate repro-ducibility for the GSI patterns of the populations exam-ined (DB2), which is shown in the Figure 2.
4. Discussion
Daily growth based on the analysis of the otolithmicrostructure, together with information from thegonads, has been widely used to obtain information onthe reproductive biology of fish species. From ichthy-oplankton studies, or based on later-stage individuals,hatching date distributions, determined by backwardcalculations of age from the date of capture in otolithreadings, is an extended methodology to determinespawning behaviour and, therefore, environmentalfactors related to it (Campana and Jones, 1992).However, the hatching date distribution obtained fromthese observations could be biased due to the lack ofindividuals at certain stages, mainly due to advectiveor dispersive processes or to mortality rates that occurduring the time between the birth of the individualsand their capture (Campana and Jones, 1992). Themajority of studies that have determined hatchingdate distributions using this methodology are basedon larvae or early juvenile (young of the year) stages,where the correction factors applied are based on theabsence or survival of individuals between successivecohorts or age classes (most of which are arbitrarilyset within the same spawning season). For example,Methot and Wetzel (2013) corrected the hatch-datedistributions by the inverse of survival rates betweenage classes, whereas others studies, focused on larvalstages, corrected the hatch distribution through theinstantaneous mortality rate determined for the lifeperiod studied (Yoklavich and Bailey, 1990; Fortier andQui˜nonez-Velazquez, 1998; Marteinsdottir et al., 2000;Qui˜nonez-Velazquez et al., 2000). Qui˜nonez-Velazquezet al. (2000) also applied a correction based on a mor-tality coe ffi cient for juvenile pacific sardine. All thesestudies are conducted within the critical life period forfish species, when mortality rates are highly significant,with strong e ff ects in the shape of the back-calculateddistributions (Campana and Jones, 1992). These au-thors a ffi rm that after the critical period, once juvenilehave attained a determinate age, in which mortalityrates have decreased to minimum values, the shape ofthe observed and expected hatch-distributions will besimilar. This is the case of our study, because althoughwe are sampling on young of the year individuals,they are already survivors of the critical early lifeperiod. Thus, we applied a constant natural mortalitycoe ffi cient acting throughout the entire life period. Moreover, once individuals attain a certain length (withis variable in age depending on the thermal historyexperienced according to the growth model used (Molt´oet al. submitted), they will be available (vulnerable)to the fishery which is, in turn, legally establishedwith a certain period in the Mediterranean Sea (Rec-ommendation GFCM / / / // / gfcm / data / reporting / dolphinfish / en / ).Once the fish is vulnerable to the fishery, the fishingmortality could act in a similar way to the naturalmortality during the critical stage in terms of thetheoretical hatch-date distribution determination.The assumption behind the utilization of a constantmortality rate ignores an age or length-specific mortal-ity, which is more likely to realistically occur than afixed mortality coe ffi cient (Campana and Jones, 1992).Despite fixed coe ffi cients have been commonly used, asin this work, variable mortality rates along age or sizewould be a more accurate mathematical approach in or-der to better obtain the theoretical hatch-date distribu-tion. Variable mortality values could be incorporated inour model as we are simulating the age and size alongthe entire life period of each individual when the birthdate is back-calculated. However, these data are di ffi cultto determinate and are not available yet for this species.This is the first time that a methodology is provided toestimate the theoretical hatch-date distribution from ju-veniles after the critical mortality stage and already in-corporated and sampled from the fishery, accounting fora constant mortality rate. Moreover, our model incorpo-rates the environmental factors with strong influence onthe shape of the size distribution due to regional thermalregimes. The thermal regimes are incorporated togetherwith the GSI information of di ff erent dolphinfish pop-ulations analysed around the globe, as well as the indi-vidual daily thermal histories, which are implicit in thegrowth model used to back-calculate the hatching-datedistribution (Molt´o et al. submitted).This work provides a methodology to recover thetheoretical hatch-date distribution from a certain co-hort. Thus, it can be used to set-up the expected hatchdates distribution under a specific thermal regime; and,due to its coupling to a growth model, used to developpopulation-based simulations of expected length distri-bution at a given date, under a given temperature andmanagement (fishing mortality rates) scenarios. Acknowledgements
This work has received funding from the EuropeanUnion’s Horizon 2020 Research and Innovation pro-gram under Grant Agreement 678193 (CERES). AO4 igure 1: Shift in hatching date distribution for a dolphinfish cohort from Balearic Islands for the year 2004. The red shaded area correspondsto the distribution of observed hatching dates and the blue area represents the distribution of theoretical hatching dates estimated by the Bayesianmethod explained in this research. The vertical bars (red and blue) indicate the mean values of each distribution.Figure 2: GSI patterns of the dolphinfish populations examined.
References
Anderson, J.J., Gurarie, E., Bracis, C., Burke, B.J., Laidre, K.L.,2013. Modeling climate change impacts on phenology andpopulation dynamics of migratory marine species. Ecolog-ical Modelling 264, 83–97. URL: https://linkinghub.elsevier.com/retrieve/pii/S030438001300166X , doi: .Campana, S.E., Jones, C.M., 1992. Analysis of Otolith Mi-crostructure, in: Stevenson, D.K., Campana, S.E. (Eds.), OtolithMicrostructure Examination and Analysis. Can. Spec. Publ. Fish.Aquat. Sci., pp. 73–100. URL: .CopeMed, II, 2016. CopeMed II-MedSudMed Technical Workshopon
Coryphaena hippurus
Fisheries in the Western-Central Mediter-ranean Malta 16- 18 March 2016. CopeMed II Technical Docu-ments No.42, GCP / INT / / SPA; GCP / INT / / EC , 1–31.Decker, F.J., 1994. Beam distributions beyond RMS. Techni-cal Report. Stanford Linear Accelerator Center, Stanford, CA.URL: https://ui.adsabs.harvard.edu/abs/1994STIN...9610487D .Dobson, S., Dodd, J.M., 1977. The roles of temperatureand photoperiod in the response of the testis of the dogfish,Scyliorhinus canicula L. to partial hypophysectomy (ventral lobec-tomy). General and Comparative Endocrinology 32, 114–115.URL: .Folkvord, A., Gundersen, G., Albretsen, J., Asplin, L.,Kaartvedt, S., Giske, J., 2015. Impact of hatch date onearly life growth and survival of Mueller’s pearlside (Mau-rolicus muelleri) larvae and life-history consequences1.Canadian Journal of Fisheries and Aquatic Sciences URL: , doi: .Fortier, L., Qui˜nonez-Velazquez, C., 1998. Dependence of sur-vival on growth in larval pollock Pollachius virens and had-dock Melanogrammus aeglefinus: A field study based on in-dividual hatchdates. Marine Ecology Progress Series 174,1–12. URL: , doi: .Furukawa, S., Ohshimo, S., Tomoe, S., Shiraishi, T., Nakatsuka,N., Kawabe, R., 2012. Age, growth, and reproductive char-acteristics of dolphinfish Coryphaena hippurus in the waters o ff west Kyushu, northern East China Sea. Fisheries Science 78,1153–1162. URL: http://link.springer.com/10.1007/s12562-012-0557-6 , doi: .Gatt, M., Dimech, M., Schembri, P.J., 2015. Age, Growth and Re-production of Coryphaena hippurus (Linnaeus, 1758) in MalteseWaters, Central Mediterranean. Mediterranean Marine Science16, 334–345. URL: https://ejournals.epublishing.ekt.gr/index.php/hcmr-med-mar-sc/article/view/12631 ,doi: . Gelman, A., Carlin, J.B., Stern, H.S., Dunson, D.B., Ve-htari, A., Rubin, D.B., 2013. Bayesian Data Analysis.3rd ed., Chapman and Hall / CRC, Boca Raton, FL. URL: .Islam, M.S., Ueno, M., Yamashita, Y., 2015. Recruit-ment in Japanese sea bass, Lateolabrax japonicas (Cuvier,1828): e ff ects of timing on spawning and larval qualityand quantity. Journal of Applied Ichthyology 31, 1064–1071. URL: https://onlinelibrary.wiley.com/doi/full/10.1111/jai.12861 , doi: .Kruschke, J.K., 2015. Chapter 8 - JAGS, in: Kruschke,J.K. (Ed.), Doing Bayesian data analysis : a tutorial withR, JAGS, and Stan. Academic Press, Boston, pp. 193–219.URL: .Lapolla, A., Buckley, L.J., 2005. Hatch date distributionsof young-of-year haddock Melanogrammus aeglefinus in theGulf of Maine / Georges Bank region: implications for re-cruitment. Marine Ecology Progress Series 290, 239–249.URL: , doi: .Lee, H.H., Maunder, M.N., Piner, K.R., Methot, R.D., 2011. Estimat-ing natural mortality within a fisheries stock assessment model:An evaluation using simulation analysis based on twelve stock as-sessments. Fisheries Research 109, 89–94. URL: http://dx.doi.org/10.1016/j.fishres.2011.01.021 , doi: .Marteinsdottir, G., Gunnarsson, B., Suthers, I.M., 2000. Spatial vari-ation in hatch date distributions and origin of pelagic juvenile codin Icelandic waters. ICES Journal of Marine Science 57, 1182–1195. URL: https://doi.org/10.1006/jmsc.2000.0880 ,doi: .Massut´ı, E., Morales-Nin, B., 1995. Seasonality and reproductionof dolphin-fish (Coryphaena hippurus) in the western Mediter-ranean. Scientia Marina 59, 357–364. URL: http://scimar.icm.csic.es/scimar/index.php/secId/6/IdArt/2782/ .Massut´ı, E., Morales-Nin, B., Moranta, J., 1999. Otolith mi-crostructure, age, and growth patterns of dolphin, Coryphaenahippurus, in the western Mediterranean. Fishery Bulletin 97,891–899. URL: https://digital.csic.es/handle/10261/55916 , doi: .McGovern, J.C., Olney, J.E., 2011. Factors a ff ecting sur-vival of early life stages and subsequent recruitment ofstriped bass on the Pamunkey River, Virginia. Cana-dian Journal of Fisheries and Aquatic Sciences 53, 1713–1726. URL: , doi: .McQuinn, I.H., 1989. Identification of Spring- and Autumn-Spawning Herring ( Clupea harengus harengus) using Ma-turity Stages Assigned from a Gonadosomatic Index Model.Canadian Journal of Fisheries and Aquatic Sciences 46, 969–980. URL: , doi: .Methot, R.D., 1983. Seasonal variation in survival of larval Engraulismordax estimated from the age distribution of juveniles. Fish-ery Bulletin 81, 741–750. URL: , doi: .Methot, R.D., Wetzel, C.R., 2013. Stock synthesis: Abiological and statistical framework for fish stock as-sessment and fishery management. Fisheries Research142, 86–99. URL: , oi: .Neuheimer, A.B., MacKenzie, B.R., 2014. Explain-ing life history variation in a changing climate acrossa species’ range. Ecology 95, 3364–3375. URL: https://esajournals.onlinelibrary.wiley.com/doi/full/10.1890/13-2370.1 , doi: .Pankhurst, N.W., Munday, P.L., 2011. E ff ects of climate change onfish reproduction and early life history stages. Marine and Fresh-water Research 62, 1015–1026. URL: , doi: .Plummer, M., 2003. JAGS: A program for analysis of Bayesian graph-ical models using Gibbs sampling, in: Proceedings of the 3rd in-ternational workshop on distributed statistical computing, Vienna,Austria.. p. 10.Plummer, M., 2008. Penalized loss functions for Bayesianmodel comparison. Biostatistics (Oxford, England) 9, 523–539. URL: https://academic.oup.com/biostatistics/article-lookup/doi/10.1093/biostatistics/kxm049 ,doi: .Poloczanska, E.S., Burrows, M.T., Brown, C.J., Garc´ıa Moli-nos, J., Halpern, B.S., Hoegh-Guldberg, O., Kappel, C.V.,Moore, P.J., Richardson, A.J., Schoeman, D.S., Syde-man, W.J., 2016. Responses of Marine Organisms toClimate Change across Oceans. Frontiers in Marine Sci-ence 3, 515. URL: http://journal.frontiersin.org/Article/10.3389/fmars.2016.00062/abstract ,doi: .P¨ortner, H.O., Peck, M., 2010. Climate change e ff ectson fishes and fisheries: towards a cause-and-e ff ect un-derstanding. Journal of Fish Biology 77, 1745–1779.URL: ,doi: .Qui˜nonez-Velazquez, C., Nevarez-Mart´ınez, M.O., Gluyas-Mill´an,M.G., 2000. Growth and hatching dates of juvenile pacificsardine Sardinops caeruleus in the Gulf of California. FisheriesResearch 48, 99–106. URL: ,doi: .R Core Team, 2019. R: A Language and Environment for Statisti-cal Computing. R Foundation for Statistical Computing. Vienna,Austria. URL: .Rajesh, K.M., Rohit, P., Abdussamad, E.M., 2016. Fishery,diet composition and reproductive biology of the dolphinfishCoryphaena hippurus (Linnaeus, 1758) o ff Karnataka, south-west coast of India. Indian Journal of Fisheries 63, 1–6.URL: http://epubs.icar.org.in/ejournal/index.php/IJF/article/view/60190 , doi: .Schwenke, K.L., Buckel, J.A., 2008. Age, growth, and re-production of dolphinfish (Coryphaena hippurus) caught o ff the coast of North Carolina. Fishery Bulletin 106, 82–92. URL: http://agris.fao.org/agris-search/search.do?recordID=AV2012089738 .Stawitz, C.C., Haltuch, M.A., Johnson, K.F., 2019. How doesgrowth misspecification a ff ect management advice derived froman integrated fisheries stock assessment model? Fisheries Re-search 213, 12–21. URL: , doi: .Then, A.Y., Hoenig, J.M., Hall, N.G., Hewitt, D.A., Jardim, H.e.E.,2014. Evaluating the predictive performance of empirical esti-mators of natural mortality rate using information on over 200fish species. ICES Journal of Marine Science 72, 82–92. URL: https://doi.org/10.1093/icesjms/fsu136 , doi: .Walker, N.D., Garc´ıa-Carreras, B., Le Quesne, W.J.F., Maxwell, D.L.,Jennings, S., 2019. A data-limited approach for estimating fish-ing mortality rates and exploitation status of diverse target andnon-target fish species impacted by mixed multispecies fisheries.ICES Journal of Marine Science 76, 824–836. URL: https://doi.org/10.1093/icesjms/fsy205 , doi: .Wu, C.C., Su, W.C., Kawasaki, T., 2001. Reproduc-tive biology of the dolphin fish Corphaena hippurus onthe east coast of Taiwan. Fisheries Science 67, 784–793. URL: ,doi: .Xie, S., Watanabe, Y., 2005. Hatch date-dependent di ff erences inearly growth and development recorded in the otolith microstruc-ture of Trachurus japonicus. Journal of Fish Biology 66, 1720–1734. URL: https://onlinelibrary.wiley.com/doi/full/10.1111/j.0022-1112.2005.00764.x , doi: .Yoklavich, M.M., Bailey, K.M., 1990. Hatching period, growth andsurvival of young walleye pollock Theragra chalcogramma as de-termined from otolith analysis. Marine Ecology Progress Se-ries 64, 13–23. URL: .Z´u˜niga Flores, M.S., Ortega-Garc´ıa, S., Rodr´ıguez-Jaramillo,M.D.C., L´opez-Mart´ınez, J., 2011. Reproductive dynam-ics of the common dolphinfish Coryphaena hippurusin thesouthern Gulf of California. Marine Biology Research 7,677–689. URL: , doi: . ppendix A. Dolphinfish juveniles’ otolith readings database Table A.1: Otolith readings of juvenile dolphinfish, including date ofcapture, for 13 Mediterranean cohorts (3 years from the Balearic Islandsand 8 years from Malta).
Area Catch Date Age FL (cm) Birthdate
Balearic Islands 09 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Balearic Islands 18 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Balearic Islands 18 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Balearic Islands 28 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Balearic Islands 30 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Balearic Islands 04 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 29 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 29 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 29 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 29 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 05 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 30 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 20 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 20 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 22 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 24 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 07 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 26 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 07 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 14 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 06 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / able A.1 continued from previous pageArea Catch Date Age FL (cm) Birthdate Malta 07 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / ppendix B. IGS databases Table B.2: Monthly population-averaged GSI for 7 populations aroundthe world (code and source): Taiwan (TAIW, Wu et al. 2001), Malta(MAL, Gatt et al. 2015) , Balearic Islands (BAL, Massut´ı and Morales-Nin 1995 and this work), North Carolina (NCAROL, Schwenke andBuckel 2008), South West India (SWINDIA, Rajesh et al. 2016), Japan(JAP, Furukawa et al. 2012) and California (CALIF, Z´u˜niga Flores et al.2011). The mean, standard deviation and delta Sea Surface Temperature(SST) are indicated.
Area Month Mean SST SD SST Delta SST Mean Latitude Mean IGS N (IGS)
TAIW 1 22.752 0.502 0.063 23.350 3.713 NATAIW 2 23.242 0.614 0.490 23.350 5.090 NATAIW 3 24.515 0.546 1.273 23.350 5.150 NATAIW 4 26.445 0.404 1.930 23.350 4.132 NATAIW 5 27.714 0.480 1.269 23.350 3.952 NATAIW 6 28.717 0.340 1.003 23.350 4.012 NATAIW 7 28.759 0.387 0.042 23.350 2.814 NATAIW 8 28.281 0.309 -0.478 23.350 3.533 NATAIW 9 26.975 0.549 -1.305 23.350 1.796 NATAIW 10 25.710 0.638 -1.265 23.350 1.617 NATAIW 11 23.774 0.341 -1.936 23.350 2.275 NATAIW 12 22.688 0.307 -1.086 23.350 1.976 NAMAL 1 15.186 0.667 -1.003 35.730 NA NAMAL 2 15.009 0.606 -0.177 35.730 NA NAMAL 3 15.927 0.535 0.918 35.730 NA NAMAL 4 18.538 0.453 2.612 35.730 NA NAMAL 5 21.830 0.748 3.291 35.730 8.656 NAMAL 6 25.547 0.589 3.718 35.730 6.035 NAMAL 7 26.472 0.631 0.925 35.730 6.436 NAMAL 8 25.885 0.524 -0.587 35.730 NA NAMAL 9 23.757 0.211 -2.128 35.730 2.739 NAMAL 10 21.009 0.921 -2.748 35.730 2.535 NAMAL 11 18.184 0.597 -2.825 35.730 1.341 NAMAL 12 16.189 0.675 -1.995 35.730 1.259 NABAL 1 13.669 0.377 -0.884 39.775 NA NABAL 2 13.755 0.417 0.086 39.775 NA NABAL 3 15.246 0.559 1.491 39.775 NA NABAL 4 17.981 0.792 2.735 39.775 NA NABAL 5 22.001 0.862 4.020 39.775 2.581 NABAL 6 25.317 0.712 3.316 39.775 13.410 NABAL 7 26.441 0.719 1.124 39.775 10.369 NABAL 8 25.059 0.386 -1.382 39.775 8.848 NABAL 9 22.831 0.646 -2.229 39.775 6.544 NABAL 10 19.207 0.893 -3.624 39.775 3.272 NABAL 11 16.152 0.634 -3.054 39.775 1.521 NABAL 12 14.553 0.517 -1.600 39.775 NA NANCAROL 1 18.394 0.689 -0.850 35.103 2.559 4NCAROL 2 18.659 1.007 0.265 35.103 NA NANCAROL 3 19.760 0.728 1.101 35.103 1.681 630 able B.2 continued from previous pageArea Month Mean SST SD SST Delta SST Mean Latitude Mean IGS N (IGS)able B.2 continued from previous pageArea Month Mean SST SD SST Delta SST Mean Latitude Mean IGS N (IGS)