Contents lists available at ScienceDirect

Fisheries Research

journal homepage: www.elsevier.com/locate/fishres

Whale shark-tuna associations, insights from a small pole-and-line fisheryfrom the mid-north Atlantic

Jorge Fontesa,b,*, Niall McGintyc, Miguel Machetea,b, Pedro Afonsoa,b

aOKEANOS-UAc University of the Azores, Horta, Azores, Dept. of Oceanography and Fisheries, 9900-862, Horta, Portugalb IMAR-Azores, Institute of Marine Research at the University of the Azores, Dept. of Oceanography and Fisheries, 9900-862, Horta, Portugalc School of Oceanography, Dalhousie University, Halifax, Canada

A R T I C L E I N F O

Handled by B. Morales-Nin

Keywords:Whale sharkTunaMegafaunaAssociationsPole-and-lineTuna fisheriesAzores

A B S T R A C T

Fisherman have known and explored the association between tuna and whale sharks for centuries. Just as in thepast, present day industrial fishers search for the presence of whale shark as cue to locate tuna schools. Thesewhale shark-tuna associations have been usually assumed to result from the same ecological drivers that governthe attraction of tunas and other fishes towards floating objects or Fish Aggregation Devices (FAD). Yet, thisassumption has never been explicitly investigated. As a result, the ecological (evolutionary) significance of thewhale shark-tuna associations and the potential impacts of fishing these associations remain unclear. Here, weset out to test a set of predictions expected under the framework of the ‘indicator-log’ and the ‘meeting-point’hypotheses, originally proposed in the context of the tuna-FAD associative behaviour, by contrasting inter-annual fishing patterns, over 16-years, across highly variable whale shark abundance periods and comparingwhale shark associated tuna schools and free swimming schools in the Azores. In addition we also analysed thetuna fishing fleet behaviour over periods of high and low whale shark abundance across the archipelago.

We found an overall south and eastward migration of fishing effort towards lower latitudes, where mostwhale sharks were reported, compared to the low whale shark occurrence period. Overall we found that thewhale shark associated school differed from free swimming schools and fishing yields were also influenced bythe presence of the whale shark. Specifically, we found that skipjack, bigeye, albacore and yellowfin tuna arecaught in higher quantities when associated with whale sharks. Whale shark associated schools were morediverse and yielded smaller fish on average than free-swimming schools, as predicted under the theoreticalframework of tuna-FAD associations. Our results highlight the possibility that whale shark behaviour may playan active role in the association dynamics, and that these associations may be mutually beneficial. We argue thatwhale sharks may represent a “meeting point” for the aggregation of small tunas and that this behaviour is morelikely to be size dependent than species oriented. Finally, we discuss the potential implications of our findings forthe management and conservation of both tunas and whale sharks.

1. Introduction

Tunas represent about 10 % of the global seafood trade, with about6 million tons landed annually (FAO, 2019). The growing global de-mand for tuna and the increasing efficiency of the fishing fleets haveresulted in the exploitation of most stocks up to its full capacity, andeven to its overexploitation in some cases (Coulter et al., 2020; Cullis-Suzuki and Pauly, 2010). Despite such large pressure on tuna stocks,our knowledge of the complex ecology of tunas is still quite incomplete,mostly due to the practical challenges of investigating these highlymigratory oceanic species. This limitation often leads to inefficient orcomplex management schemes for transnational tuna stocks, usually

shared among multiple national exclusive economic zones and highseas (Allen et al., 2010).

A particularly striking knowledge gap is the behavioural ecology oftuna and their interactions with other oceanic species, such as whaleshark Rhincodon typus Smith 1828, the largest fish on earth and othersensitive megafauna (Escalle et al., 2019). Fisherman have known andexplored the association between tuna and whale sharks for centuries.Just as in the past, present day industrial fishers search for the presenceof whale sharks and other megafauna as cues to locate tuna schools(Fromentin et al., 2014; Majkowski, 2007). Although the bycatch ofwhale sharks in the industrial purse seine tuna fisheries is small, withdirect mortality estimates ranging from none in the eastern Atlantic

https://doi.org/10.1016/j.fishres.2020.105598Received 31 May 2019; Received in revised form 2 April 2020; Accepted 6 April 2020

⁎ Corresponding author.E-mail address: jorge.mr.fontes@uac.pt (J. Fontes).

Fisheries Research 229 (2020) 105598

0165-7836/ © 2020 Elsevier B.V. All rights reserved.

T

(Escalle et al., 2018) to 1% in the Indian ocean (Capietto et al., 2014),some tuna regional fisheries management organizations (e.g., IndianOcean Tuna Commission (IOTC), Western and Central Pacific FisheriesCommission (WCPFC), and Inter-American Tropical Tuna Commission(IATTC)),imposed a moratoria on the intentional net setting on whalesharks (Escalle et al., 2018).

Despite the growing interest in assessing the impact of fishingpractices on these vulnerable and emblematic megafauna, quantitativestudies on the catch composition in whale shark associated fishingevents are extremely scarce, in contrast with the volume of informationavailable for free school and FAD associated fishing (Andrzejaczeket al., 2019). The few studies that have addressed whale shark-tunaassociations have predominately focussed on the bycatch and survivalof the whale sharks (e.g. Andrzejaczek et al., 2019; Capietto et al.,2014; Román et al., 2018; Romanov, 2002).

Whale shark-tuna associations have been usually assumed to bemotivated by the same forces driving tunas and other fishes to associatewith floating objects or Fish Aggregation Devices (FAD) (Freon andDagorn, 2000), an assumption that has supported its classificationunder the same category as vessel and log associations in the purse-seine tropical fisheries (Pallarés and Petit, 1998; Pianet et al., 2000).The theory and empirical evidence to explain the associative behaviourof tuna with floating objects was comprehensively reviewed, and es-sentially consider the combined advantages of the ‘indicator-log’ andthe ‘meeting-point’ hypotheses as the most credible drivers of this be-haviour (Castro et al., 2002; Freon and Dagorn, 2000). The indicator-log hypothesis postulates that tunas may use natural floating objects asindicators of productive areas, either because most natural floatingobjects originate in rich areas (i.e., river mouth, mangrove swamps) andremain within these rich bodies of water, or because they aggregate inrich frontal zones. The meeting point hypothesis assumes that in-dividual tuna will increase their encounter rate with other isolated orschooling individuals by associating with floating objects, thus poten-tially benefiting from the evolutionary advantages of schooling: re-duced risk of predation (via a dilution effect), faster location of food,increased time for feeding, increased effective sampling and (social)information transfer, and opportunity for learning by social facilitation(Fréon and Misund, 1999; Pitcher and Parrish, 1993). Freon and Dagorn(2000) extend the meeting point hypothesis to explain associations oftuna with drifting objects, anchored FADs and slowly moving marinemegafauna, such as whale sharks. In light of these arguments, it isreasonable to expect that, as in FADs, whale shark associated tunaschools will be composed of smaller fish (Hallier and Gaertner, 2008)and be more diverse (Andrzejaczek et al., 2019; Escalle et al., 2016)than free-swimming schools. However, there are no specific empiricalor experimental studies to date designed to evaluate the dynamics ofthe whale shark-tuna interaction per se.

In the Azores archipelago, mid-north Atlantic, tunas represent morethan half of the total seafood landed (350,000 tons from 1950 to 2010 -Pham et al., 2013). Tuna fishing in this oceanic region has remainedartisanal, relying exclusively on pole-and-line and hand-line fishingwith live bait carried out by the local small to medium sized (up to 30mlong) boats (Morato et al., 2008b). Skipjack (Katsuwonus pelamis) andbigeye (Thunnus obesus) tunas are the most important species, with al-bacore (T. alalunga), yellowfin (T. albacares) and bluefin tunas (T.thynnus) caught in much smaller quantities. The wider Azores regionrepresents the fringing (northern limit) habitat for the three tropicaltuna, skipjack, bigeye and yellowfin, and the northern distribution andthermal limit of whale sharks in the Atlantic. Whale sharks have beenknown to occur in the region for some time but the first official recordsdate from the mid-nineties (Santos et al., 1995, 1997). Historically,whale shark sightings in the Azores have been typically low or null,until the dramatic increase observed in 2008, attributed to a con-comitant rise in average water temperature (Afonso et al., 2014). As inother tuna fisheries, the local fleet has also learned to opportunisticallyuse whale shark-tuna associations to locate and fish tunas. Data from

the Azorean Fisheries Observer program (POPA) shows that, whenwhale sharks are present in the region, nearly 100 % of the whale sharksightings (over 2000 from 1998 to 2013) were associated with tunaschools and resulted in the catch of tunas (Afonso et al., 2014). Thereare no records of bycatch or incidental hooking of whale sharks in thishighly selective fishery (Afonso et al., 2014; Silva et al., 2002). How-ever, as for industrial tuna fisheries, there is no information regardingthe potential impact of artisanal pole-and-line fishing on the vulner-ability of both tunas and whale sharks.

In summary, our current knowledge about the nature of whaleshark-tuna associations, their ecological (evolutionary) significance anddrivers, as well as the potential impacts from fisheries, remains poor. Inorder to address this gap, we first need to understand the basic dy-namics of whale shark-tuna associations and the fine-scale interactionswith fisheries. Here we set out to analyse the extensive POPA observerdatabase (from 1998 to 2013) for the pole-and-line tuna fishery in theAzores to investigate how the presence of whale sharks may impactfisheries yields and the vulnerability of associated tunas. Specifically,we sought to test the following hypothesis: i) is the spatial distributionof fishing events related to the presence of whale sharks? ii) does whaleshark-associated fishing events equate to higher fishing yields? iii) arewhale shark-associated tunas smaller than free-swimming tuna? iv) arewhale shark-associated tuna schools more diverse than free-swimmingschools?

2. Materials and methods

2.1. Study site

The Azores archipelago is located along the northern mid-Atlanticridge and is composed of nine islands broadly grouped into a western, acentral and an eastern island groups (Fig. 1). The region is character-ized by its isolation form other land masses, by the presence of a highnumber seamounts of varying heights and summit depths (Morato et al.,2008a), and by an oceanographic regime that broadly sets the region’secotone position between warm temperate and subtropical characters(Caldeira and Reis, 2017).

2.2. Fisheries related data

Whale shark sightings, which resulted in whale shark associatedfishing events in almost 100 % of cases, were extracted from the da-tabase of the Azorean pole-and-line tuna fisheries observer programme(POPA). The program deploys trained observers since 1998 in about 50% of the fleet during the entire tuna (warm) season, which typicallyextends from May to November. Boats actively search for tuna at thesurface to fish during the day, and will come into port every 2–8 days,depending on the amount of fish caught. Observers record georefer-enced data of all fishing events, including duration of fishing events andcatch (number of individuals, species and estimated size/weight), aswell as other scientifically relevant information such as the sighting ofassociated or non-associated species including turtles, seabirds, marinemammals and whale sharks. An observer will consider a whale sharkassociated fishing event if a whale shark is spotted within 15m of theboat during a fishing event (Afonso et al., 2014). The average speed ofthe tuna vessels when searching is approximately eight knots(∼15 km h−1), and the vessel will typically cover an average of 1300linear km over a week. Whale sharks are known to spend the majority(80 %) of their daytime very close to the surface (e.g. Berumen et al.,2014; Brunnschweiler and Sims, 2011; Eckert and Stewart, 2001).Therefore, we can reasonably assume that the probability of daytimewhale shark detections by observers should be high (Afonso et al.,2014). Yet, it is possible that the observers may underestimate theabundance of whale sharks and tuna, simply because they are notwithin detectable range and depth 100 % of the time. Nevertheless, thisbias should not affect our results as they do not depend on the absolute

J. Fontes, et al. Fisheries Research 229 (2020) 105598

2

density and this bias should be similar across the study period. Whalesharks were frequently observed by dive operators from Sta. Maria is-land between 2008 and 2013. Yet, no systematic records were madethus precluding the use of this additional information source. For thisstudy, we considered data from all the tuna species fished in the Azoresexcept Atlantic bluefin tuna as it was only caught in 59 fishing events.Although captured in much lower numbers, the albacore and yellowfinwere captured during enough whale shark associated fishing events tobe modelled. Tuna catches were derived from direct observer counts ofthe number of individuals per species fished during a given event, whilefish size and weight were derived from the captain´s estimations ofaverage fish size and weight per species for each fishing event. As aresult, the catch per unit effort was calculated dividing the number ofindividuals collected over the duration of each fishing event (CPUE;number of individuals min−1). We calculated the CPUE for each speciesat each event.

2.3. Statistical analyses

In the Azores, the whale shark occurrence patterns were quite un-predictable since 1998. Whale shark associated fishing events were veryfew until 2007, peaked in 2008, and remained relatively high comparedto the previous period (up to 9% vs. ∼0.1 % of all events) (Table 1, Sup1). Thus, we investigated the effect of whale shark presence on thelocation of fishing events (hypothesis i. is the spatial distribution offishing events related to the presence of whale sharks?) by comparingthe average annual location of fishing events and average whale sharksighting locations. We computed the annual 75 % Kernel UtilisationDistribution (75 % KUD) for each tuna species. We also computed thecenter of gravity for all tuna species fishing events on free swimmingschools and whale shark associated to illustrate any spatial changes in

the fishing effort for each year. For whale shark associated fishingevents, the COG is simply the average latitude and longitude location ofindividual sighting for each year. We use the average position of whaleshark associated fishing events between 2008−2013 as a referencepoint and compute the distance of each annual COG from this location.We hypothesised that if fishing effort distribution was related to theoccurrence of whale sharks then the distances in annual COG after 2007would be closer to the reference point than earlier years for each tunaspecies.

To test hypotheses ii. (does whale shark-associated fishing eventsequate to higher CPUE?), hypothesis iii. (are whale shark-associatedtunas smaller than free-swimming tuna?) and hypothesis iv. (are whaleshark-associated tuna schools more diverse than free-swimmingschools?) we focused our analysis on the 2007–2013 period, from Junethrough September, when whale sharks were most abundant and eachfishing event was classified as a free swimming (FSS) or whale shark-associated tuna school. An exploratory analysis revealed that the re-lationships between the response variables and several predictor vari-ables were non-linear, so we chose a binomial generalised additivemodel (GAM) for these analyses. In order to analyse the effect of whaleshark presence on tuna CPUE and average size, we set the statisticalmodel response variable as the binomial classifier (FSS or whale shark-associated). We also included distance of each fishing event to thenearest seamount and the sea surface temperature (SST) in these modelsas those environmental factors seem important for the distribution ofwhale sharks in the Azores region (Afonso et al., 2014). GAMs allow formore complex relationships to be modelled and use smoothing splinesin place of linear coefficients as covariates. To reduce potential over-fitting we limited the smoothing splines to a maximum of 4 estimateddegrees of freedom. Variable selection followed the procedure outlinedby (Marra and Radice, 2013) where a double penalty approach is usedin combination with a maximum likelihood estimator of the regressionsplines. The extra penalty approach provides a robust way of removingany variables that do not contribute significantly to the model and re-moves any bias through stepwise selection measures (Whittinghamet al., 2006).

Hypothesis iv. (are whale shark-associated tuna schools more di-verse than free-swimming schools?) was tested by means of comparingthe tuna diversity fished at FSS and whale shark-associated. We used anANOVA to model the differences in tuna species diversity of FSS vs.whale shark-associated schools. We used the number of tuna speciescaptured at each fishing event as measure of diversity, ranging from 1(monospecific catch) to 4 (all species present in the catch). All statis-tical analysis was done using the R programming environment (R-Core-Team, 2013)

Fig. 1. The location of the Azores archipelagoin the North Atlantic (inset). The location ofthe total number of fishing events betweenMay and October from 1998-2013(n=20,079; blue dot), whale shark(Rhincodon typus) associated events (n=743;red circle). Large seamounts are represented byblack triangles (For interpretation of the re-ferences to colour in this figure legend, thereader is referred to the web version of thisarticle.).

Table 1The total number of fishing events and the number of times each tuna speciesand whale shark were found at each event across all years (1998-2013). Alsoshown is the total number of free swimming and whale shark (WS) associatedevents after 2008 and the number of time each tuna species were found at each.

Total Free swimming WS associated(1998−2013) (2008−2013) (2008−2013)

All events 20,078 5777 729K. pelamis 13,274 3423 689T. obesus 7781 3160 317T. alalunga 1027 520 11T. albacares 319 152 18Whale shark 743

J. Fontes, et al. Fisheries Research 229 (2020) 105598

3

3. Results

We analysed 20,079 fishing events recorded by observers between1998 and 2013, 753 of these events were associated with whale sharks.Most fishing events were recorded around the islands and large sea-mounts south of the central group. Annual tuna landing in the Azoreshave been highly variable, for all three species, across the entire periodof this study. We also found no correlation between annual tuna land-ings for any of the species, and whale shark sightings from 1998 to 2013(Sup 1). Between 1998 and 2007, most fishing events were recordedaround the central islands. However, after 2007, both the COG and 75% KUD for both skip jack and bigeye tuna shifted south-eastward, closerto Santa Maria island, where most whale shark associated sets wereconcentrated. The exception was the year 2012 (Fig. 2a and b respec-tively) when most skipjack and bigeye were fished south of the centralgroup. This spatial shift in fishing events and whale shark associatedfishing events was coincident with the westward displacement of the22 °C isotherm. Albacore and yellowfin tuna, although caught in sub-stantially lower numbers, show an identical trend although less robust(Sup 2).

A more detailed analysis showed that the distance between annualCOG for skipjack (ANOVA, F1,13= 8.23, p=0.013) and bigeye(ANOVA, F1,13= 17.51, p=0.001) tuna and the locations of whaleshark became significantly smaller on the period following 2007 (ex-cept 2012) (Fig. 3). The annual distances changed from 350−400 km to∼120 km on the post 2007 period. We also found a small (< 50 km)distance between the COG averaged across all years of both tuna spe-cies. Changes in the COG for albacore and yellowfin between the twoperiods were not significant.

Whale shark presence was positively correlated with CPUE for allfour tuna species, suggesting that larger yields are expected for whaleshark-associated fishing events. The strongest relationships were foundfor the two most abundant tuna species (skipjack and bigeye), for whichpositive residuals were found when CPUEs were above 40–50 in-dividuals min−1 (Fig. 4a-b). For albacore and yellowfin tuna, positiveresiduals were only found for CPUEs higher than 150 individuals min−1

(Fig. 4c-d).We also found the mean individual size of whale shark-associated

tuna to be significantly smaller than free swimming tunas, for skipjack(4 cm smaller), bigeye (17 cm) and albacore tuna (20 cm) (Fig. 5 a–c).

There was no significant difference in the mean size of free-swim-ming and whale shark-associated yellowfin tuna. The deviance ex-plained by the CPUE and average size models was 35.3 % and 32.5 %respectively (Table 2). Distance to seamount and SST both came out as

significant variables in both models (hypothesis ii and iii), showing thatwhale sharks were found at intermediate distances (∼20−60 km) fromthe summit of larger seamounts and at temperatures warmer than 22°(Fig. 6).

Species diversity was significantly greater in whale shark-associatedevents compared with FSS events (mean richness 1.36 vs 1.15, F1,6469= 64.6, p < 0.001) (Fig. 7).

4. Discussion

We analysed sixteen years of fisheries observer data to investigatethe fishing patterns, catch composition and CPUE of four importanttunas in the Azorean pole-and-lime tuna fishery. Our goal was to in-vestigate how the presence of whale sharks may influence the spatialpatterns of tuna fishing in the archipelago, CPUE and catch composi-tion. Two contrasting periods of whale shark abundance emerge fromthis dataset: a period of sporadic whale shark associated fishing events,from 1998 to 2007 (0–2 events per year), followed by a dramatic in-crease in annual whale shark-associated fishing events (2008−2013)(Afonso et al., 2014).

Fig. 2. The 75 % kernel utilisation distribution of the a) skipjack (Katsuwonus pelamis) and b) bigeye tuna (Thunnus obesus) tuna catches pre 2007 (shaded red) andafter 2007 (shaded blue). Also shown is the annual centre of gravity of a) skipjack and b) bigeye catches, for the years 98-07 (red dot) and 08-13 (blue dot) and theannual centre of gravity (black triangle) of whale shark (Rhincodon typus) associated fishing events between 2008 and 2013 with the mean centre of gravity of allyears (large triangle) also shown. In both panels the mean position of the 22 °C isotherm is shown for the periods between 98-07 (dashed grey line) and 08 - 13 (solidgrey line) (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).

Fig. 3. The distance of each annual centre of gravity from the mean whale shark(Rhincodon typus) associated fishing events centre of gravity for skipjack(Katsuwonus pelamis) (filled circle; 2012 – filled square) and bigeye (Thunnusobesus) (hollow circle; 2012 – hollow square) tuna and the 3 year running meanaveraged across both species.

J. Fontes, et al. Fisheries Research 229 (2020) 105598

4

The tuna fishing boats in the Azores typically search for tuna acrossthe entire region’s EEZ and even beyond, but will often converge tolocations where successful fishing events have recently been reported,including those associated with whale sharks (Afonso et al., 2014).Given that nearly all whale sharks sighted were reported to be asso-ciated with tuna and that most whale sharks were found southeast ofthe central group, it’s not surprising that both the 75 % KUD and COGfor tuna fishing events during the high whale shark abundance period(2008−2013) shifted south relative to the 1998–2007 period. Even in2012, when whale shark–associated fishing events decreased, there wasa notable westward shift in the tuna fleet consistent with the shift inwhale shark COG towards the Princess Alice seamount region. The in-formation about the presence of whale sharks in a given area is com-monly spread across the fleet, thus potentially favouring an overlapbetween whale shark rich areas and the fishing fleet. Reports of whalesharks presence at a given area are likely to influence captains’ deci-sions and fishing patterns, which likely resulted in the observed overlapbetween fishers and whale sharks.

Given that this is a fringing seasonal habitat for both the tropicaltunas and whale sharks, it’s thus worth posing the question ‘who fol-lows who’, or ‘are tunas following whale sharks vs are the whale sharksfollowing the tuna? It may be the case that tuna and whale sharkconverge to this high latitude in favourable (including high tempera-ture) years, possibly responding to the same environmental cues (e.g.water temperature, food availability, etc.). The correlation betweenwhale shark abundance in the Azores and water temperature reportedby Afonso et al. (2014) suggests this is the case for the whale sharks.Although the distribution of tuna and whale sharks within the Azores iscomplex and difficult to predict, some marine predators like skipjackand bigeye tuna, the common dolphin (Delphinus delphis) and Cory’sshearwater (Calonectris diomedea borealis) are more likely to occur inthe proximity of seamounts’ summits shallower than 400m (Morato

et al., 2008b). Afonso et al. (2014) also reported that whale shark oc-currence within this region were more likely closer to the seamounts,coinciding with higher chlorophyll-a biomass, probably associated toincreased feeding opportunities. In fact, 63 % of all large seamounts arelocated south of the central group (Morato e al. 2008a), where mostwhale sharks were reported and where the COG of the two mostabundant tuna lay following 2007. It is thus reasonable to assume that,the distribution patterns of tunas in the Azores is likely influenced by acombination of seamount landscape, environmental cues and whaleshark habitat use patterns, as shown by the clear southward shift oftuna 75 % KUD and COG (skipjack and bigeye) after 2007. Major shiftsin tuna distribution and fishing patterns have been previously docu-mented in the tropics where most industrial fisheries operate. In somecase, these shifts have been attributed to the massive deployment ofFADs that displaced the tunas away from traditional areas of log (nat-ural FADs) fishing (Freon and Dagorn, 2000).

Although the motivation and advantages of whale shark-tuna as-sociations remain elusive, a few hypotheses have been put forward(Reviewed in Freon and Dagorn, 2000). Among these propositions, the“meeting point” and “indicator-FAD” hypotheses seem more credible(Hallier and Gaertner, 2008). Under this theoretical framework, tunasare attracted to FADs and marine megafauna in order to facilitate theformation of large schools, towards the same evolutionary advantagesof schooling (Castro et al., 2002) and food finding (Freon and Dagorn,2000). The distribution patterns of tunas in the Azores during the highwhale shark abundance period seem to support the view that FADs andwhale sharks may have a similar ecological effect on tunas, as suggestedby Pianet et al. (2000). Our finding that whale shark-associated schoolswere typically composed of smaller individuals (except yellowfin) andtypically harboured more diverse tuna schools relative to free-swim-ming schools is also in line with this hypothesis. Moreover, a recentstudy by Escalle et al. (2019) also found that purse seine sets around

Fig. 4. The partial residual plots of the four tuna species showing the log10 CPUE individuals min−1 at each fishing event (black line=mean; shaded area=95 % CI)based on the probability being associated with a whale shark (Rhincodon typus): a) Skipjack (Katsuwonus pelamis), b) Bigeye (Thunnus obesus), c) Albacore (Thunnusalalunga), d) Yellowfin (Thunnus albacares).

J. Fontes, et al. Fisheries Research 229 (2020) 105598

5

FADs and whale sharks, in the Atlantic, tend to be more diverse andproduce average smaller size catch of the three major tuna species(skipjack, bigeye and yellowfin tuna), compared to free-swimmingschool sets, especially due to the capture of large yellowfin tuna.

It is not surprising that smaller tunas, which are typically morevulnerable to predators, seem to be more likely to associate with FADsor whale sharks as this behaviour is expected reduce predation riskthrough the dilution effect (Castro et al., 2002; Hoare et al., 2000;Pitcher and Parrish, 1993). On the other hand, the higher tuna diversitytypically associated with whale sharks suggests that whale sharks mayact as a “meeting point” for the aggregation of small tunas and thisbehaviour is more likely to be size dependent than species oriented. It isexpected that different species may gather in a single school to benefitfrom the advantages of larger schools (Fréon and Misund, 1999; Pitcherand Parrish, 1993).

While small and immature fishes associated to FAD and whalesharks may benefit from reduced predation risk, i.e. decreased naturalmortality, our results suggest they may be more vulnerable to fishingcompared to free-swimming schools. Similar patterns have been re-ported for example, in the eastern Atlantic Ocean, where FAD asso-ciated sets typically generate more than twice the catch than freeswimming schools, even though the average fish size of FAD associatedfish is smaller (Fonteneau et al., 2000). Escalle et al. (2019) also re-ported that free school sets had the lowest set success rate (more thanone ton of catch) compared to whale shark sets and FAD sets. Ulti-mately, it can be argued that fishing whale shark-associated schools isless sustainable than fishing free swimming schools given the higherproportion of immature fish captured with apparently higher captur-ability.

Fig. 5. The partial residual plots of the four tuna species showing the mean size of the individuals at each fishing event (black line=mean; shaded area=95 % CI)based on the probability of being associated with a whale shark (Rhincodon typus). The dashed line shows the y-intercept= 0 and the rugplot shows the sizecategories used for each species: a) Skipjack (Katsuwonus pelamis), b) Bigeye (Thunnus obesus), c) Albacore (Thunnus alalunga), d) Yellowfin (Thunnus albacares).

Table 2GAM results exploring the effect of whale sharks on the CPUE and average sizeof individuals for each of the four tuna species. In addition to the the four tunaspecies, each model included two environmental variables, the distance toseamounts and SST. For each term the estimated degrees of freedom (Est. df.)and Chi squared statistic (Chi sq.) is shown. Significance is denoted as: ***p < 0.001, ** p < 0.01, * p < 0.05, ns= not significant.

Response: WSp/a ∼ CPUE

s(term, k= 4) Est df. Chi sq.

Log10(Skipjack CPUE) 4.0 111.1***Log10(Bigeye CPUE) 1.9 98.4***Log10(Albacore CPUE) 2.2 18.2***Log10(Yellowfin CPUE) 2.4 17.9**Distance to Seamount 1.9 64.25***SST 2.3 222.1***Dev. Explained=35.3 % n=6471

Response: WSp/a ∼ mean size cm.

s(term, k= 4) Est df. Chi sq.

Skipjack cm. 1.8 13.73**Bigeye cm. 2.5 44.0***Albacore cm. 1.8 20.5***Yellowfin cm. 0.1 0ns

Distance to Seamount 1.3 75.6***SST 1.9 229.8***Dev. Explained=32.5 % n=6471

J. Fontes, et al. Fisheries Research 229 (2020) 105598

6

Even if some parallel can be found between FAD and whale shark-associated schools, one fundamental difference remains, whale sharksare not objects passively drifting on the surface of the ocean. In fact,whale shark are exactly the opposite: they are complex organisms withacute sensory capabilities and complex behaviour, performing transo-ceanic migrations, dynamic diel vertical behaviour, capable of extremedives with complex feeding ecology (Borrell et al., 2011;Brunnschweiler et al., 2009; Cárdenas-Palomo et al., 2018; Rohneret al., 2018; Sampaio et al., 2018).

The whale shark is generally known as a filter-feeder mainly tar-geting zooplankton and micronekton, including fish spawn (de la ParraVenegas et al., 2011; Meekan et al., 2009; Rowat and Brooks, 2012;Sampaio et al., 2018). Yet, there are increasingly more evidence tosuggest that whale sharks consume more fish than previously antici-pated (Boldrocchi and Bettinetti, 2019; Borrell et al., 2011; Cárdenas-Palomo et al., 2018; Sampaio et al., 2018). Nevertheless, it is still un-clear how they are able to prey large quantities of small highly mobilefish. We speculate that the association with tunas may facilitate accessto this abundant resource as tunas force the formation of compact bait-balls, making it accessible to slow moving whale sharks (e.g. https://www.youtube.com/watch?v=_B8qiqeDrI0). This feeding interactionmay represent an important benefit for whale sharks, particularly whenand where baitfish is the only staple available. This is arguable the casein the Azores, during the summer months, when whale shark aggregate,

primary production is lowest (Woods and Barkmann, 1995) and zoo-plankton concentrations decrease to its minimum (Carmo et al., 2013).While it could also be argued that whale sharks and tunas are com-peting for the same prey, it is also safe to say that this interaction enablewhale sharks to explore a resource that otherwise would not be ac-cessible.

Recognising that whale sharks are vulnerable, ecologically im-portant and emblematic, the Western and Central Pacific FisheriesCommission and the Indian Ocean Tuna Commission banned the settingof nets around whale sharks as a precautionary measure to mitigateboth lethal and sub-lethal impacts of bycatch (Capietto et al., 2014). Incontrast, similar measures are still lacking for the Atlantic, even ifwhale shark is considered endangered (IUCN red-list). There are norestrictions to pole-and-line fishing on whale shark-associated schools,possibly because the current perception is that this artisanal fisherydoes not represent a direct threat to the whale sharks since bycatch isimpossible and sub-lethal effects are not considered. However, if theseassociations are indeed ecologically meaningful, i.e., advantageous fortunas and whale sharks, then targeting these associations may result inunknown detrimental effects on whale sharks over the long term.

This 16 year data set was uniquely placed at a time when whalesharks went from a rare visitor to a regular annual visitor to this region.It presented a valuable opportunity to investigate, for the first time, theinfluence of the endangered whale shark on pole and line tuna fisherydynamics and the vulnerability of the associated tuna species. We foundevidence that suggest a link between the spatial dynamics of whalesharks and fishing fleet behaviour in the region, either directly throughinterspecific interactions and/or indirectly by responding to the sameenvironmental cues (e.g. water temperature). On the other hand, thepresence of whale shark is associated with a significant change thecatch composition and yields compared to free swimming school fishingevents. This suggests that fishing whale shark-associated schools ispotentially less sustainable than free swimming schools since whaleshark-associations typically harbour higher proportion of immaturefishes and are more vulnerable to fishing mortality. We argue that fu-ture investigations should clarify the relevance of whale shark-tunaassociations and the implications of targeting whale shark-tuna -asso-ciations.

Ethical approval

This article does not contain any studies with animals performed byany of the authors.

Fig. 6. The partial residual plots of the environmental variables used for the CPUE models (Fig. 4) showing the probability of being associated with a whale shark(Rhincodon typus) at each fishing event (black line=mean; shaded area=95 % CI): a) distance from seamount, b) SST. The plots from the mean size of individuals(Fig. 6) show a similar trend (See Supplementary 3).

Fig. 7. Species richness (number of tuna species captured at each fishing event)from whale shark (Rhincodon typus) associated fishing events and free swimingschools fishing events. Bars represent standard error.

J. Fontes, et al. Fisheries Research 229 (2020) 105598

7

Funding

This study was supported by Fundação para a Ciência e Tecnologia(FCT), through the strategic projects UID/MAR/04292/2013 granted toMARE, UIDB/05634/2020 granted to Okeanos-UAc and the researchproject PTDC/BIA-BMA/32204/2017 IslandShark. JF and PA weresupported by Fundo Regional para a Ciência e Tecnologia (DRCT) grantM3.1.a/F/062/2016 and FCT grant IF/01640/2015, respectively.

CRediT authorship contribution statement

Jorge Fontes: Conceptualization, Investigation, Methodology,Writing - review & editing, Data curation, Formal analysis, Fundingacquisition, Writing - original draft. Niall McGinty: Conceptualization,Investigation, Methodology, Writing - review & editing, Data curation,Formal analysis, Writing - original draft. Miguel Machete:Conceptualization, Investigation, Methodology, Writing - review &editing, Data curation, Project administration. Pedro Afonso:Conceptualization, Investigation, Methodology, Writing - review &editing, Funding acquisition, Writing - original draft.

Declaration of Competing Interest

Authors declare that there are no conflicts of interest.

Acknowledgements

We thank the tuna boat crews many and observers that collected thedata, and the anonymous reviewers for improvements in the manu-script. This study had the support of Fundação para a Ciência eTecnologia (FCT), through the strategic project UID/MAR/04292/2013Granted to MARE, UIDB/05634/2020 grnated to Okeanos-UAc andresearch project PTDC/BIA-BMA/32204/2017IslandShark. JF and PAwere supported by Fundo Regional para a Ciência e Tecnologia (DRCT)grant M3.1.a/F/062/2016 and FCT grant IF/01640/2015, respectively.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in theonline version, at doi:https://doi.org/10.1016/j.fishres.2020.105598.

References

Afonso, P., McGinty, N., Machete, M., 2014. Dynamics of whale shark occurrence at theirfringe oceanic habitat. PLoS One 9.

Allen, R., Joseph, J.A., Squires, D., 2010. Conservation and Management of TransnationalTuna Fisheries. John Wiley & Sons.

Andrzejaczek, S., Gleiss, A.C., Pattiaratchi, C.B., Meekan, M.G., 2019. Patterns and driversof vertical movements of the large fishes of the epipelagic. Rev. Fish. Biol. Fish. 29,335–354.

Berumen, M.L., Braun, C.D., Cochran, J.E., Skomal, G.B., Thorrold, S.R., 2014. Movementpatterns of juvenile whale sharks tagged at an aggregation site in the Red Sea. PLoSOne 9 e103536.

Boldrocchi, G., Bettinetti, R., 2019. Whale shark foraging on baitfish off Djibouti. Mar.Biodivers. 49, 2013–2016.

Borrell, A., Aguilar, A., Gazo, M., Kumarran, R.P., Cardona, L., 2011. Stable isotopeprofiles in whale shark (Rhincodon typus) suggest segregation and dissimilarities inthe diet depending on sex and size. Environ. Biol. Fishes 92, 59–567.

Brunnschweiler, J.M., Sims, D.W., 2011. Diel oscillations in whale shark vertical move-ments associated with meso-and bathypelagic diving. Am. Fish. Soc. Symp.

Brunnschweiler, J.M., Baensch, H., Pierce, S., Sims, D., 2009. Deep‐diving behaviour of awhale shark Rhincodon typus during long‐distance movement in the western IndianOcean. J. Fish Biol. 74, 706–714.

Caldeira, R., Reis, J.C., 2017. The azores confluence zone. Front. Mar. Sci. 4 (37).Capietto, A., Escalle, L., Chavance, P., Dubroca, L., Molina, A.D., Murua, H., Floch, L.,

Damiana, A., Rowat, D., Merigot, B., 2014. Mortality of marine megafauna inducedby fisheries: insights from the whale shark, the world’s largest fish. Biol. Conserv.174, 147–151.

Cárdenas-Palomo, N., Noreña-Barroso, E., Herrera-Silveira, J., Galván-Magaña, F.,Hacohen-Domené, A., 2018. Feeding habits of the whale shark (Rhincodon typus)inferred by fatty acid profiles in the northern Mexican Caribbean. Environ. Biol.Fishes 101, 1599–1612.

Carmo, V., Santos, M., Menezes, G.M., Loureiro, C.M., Lambardi, P., Martins, A., 2013.Variability of zooplankton communities at Condor seamount and surrounding areas,Azores (NE Atlantic). Deep. Sea Res. Part II Top. Stud. Oceanogr. 98, 63–74.

Castro, J.J., Santiago, J.A., Santana-Ortega, A.T., 2002. A general theory on fish ag-gregation to floating objects: an alternative to the meeting point hypothesis. Rev. FishBiol. Fish. 11, 255–277.

Coulter, A., Cashion, T., Cisneros-Montemayor, A.M., Popov, S., Tsui, G., Le Manach, F.,Schiller, L., Palomares, M.L.D., Zeller, D., Pauly, D., 2020. Using harmonized his-torical catch data to infer the expansion of global tuna fisheries. Fish. Res. 221(105379).

Cullis-Suzuki, S., Pauly, D., 2010. Failing the high seas: a global evaluation of regionalfisheries management organizations. Mar. Policy 34, 1036–1042.

de la Parra Venegas, R., Hueter, R., Cano, J.G., Tyminski, J., Remolina, J.G., Maslanka,M., Ormos, A., Weigt, L., Carlson, B., Dove, A., 2011. An unprecedented aggregationof whale sharks, Rhincodon typus, in Mexican coastal waters of the Caribbean Sea.PLoS One 6 e18994.

Eckert, S.A., Stewart, B.S., 2001. Telemetry and satellite tracking of whale sharks,Rhincodon typus, in the Sea of Cortez, Mexico, and the north Pacific Ocean. Environ.Biol. Fishes 60, 299–308.

Escalle, L., Gaertner, D., Chavance, P., De Molina, A.D., Ariz, J., Merigot, B., 2016.Consequences of fishing moratoria on catch and bycatch: the case of tropical tunapurse-seiners and whale and whale shark associated sets. Biodivers. Conserv. 25,1637–1659.

Escalle, L., Amandé, J., Filmalter, J., Forget, F., Gaertner, D., Dagorn, L., Mérigot, B.,2018. Update on post-release survival of tagged whale shark encircled by tuna purse-seiner. Collect. Vol. Sci. Pap. ICCAT 74, 3671–3678.

Escalle, L., Gaertner, D., Chavance, P., Murua, H., Simier, M., Pascual-Alayón, P.J.,Ménard, F., Ruiz, J., Abascal, F., Mérigot, B., 2019. Catch and bycatch captured bytropical tuna purse-seine fishery in whale and whale shark associated sets: compar-ison with free school and FAD sets. Biodivers. Conserv. 28, 467–499.

FAO, 2019. Fishery Statistical Collections.Fonteneau, A., Ariz, J., Gaertner, D., Nordstrom, V., Pallares, P., 2000. Observed changes

in the species composition of tuna schools in the Gulf of Guinea between 1981 and1999, in relation with the Fish Aggregating Device fishery. Aquat. Living Resour. 13,253–257.

Freon, P., Dagorn, L., 2000. Review of fish associative behaviour: toward a generalisationof the meeting point hypothesis. Rev. Fish Biol. Fish. 10, 183–207.

Fréon, P., Misund, O.A., 1999. Dynamics of Pelagic Fish Distribution and Behaviour:Effects on Fisheries and Stock Assessment. Blackwell Science, Oxford.

Fromentin, J.-M., Bonhommeau, S., Arrizabalaga, H., Kell, L.T., 2014. The spectre ofuncertainty in management of exploited fish stocks: the illustrative case of Atlanticbluefin tuna. Mar. Policy 47, 8–14.

Hallier, J.-P., Gaertner, D., 2008. Drifting fish aggregation devices could act as an eco-logical trap for tropical tuna species. Mar. Ecol. Prog. Ser. 353, 255–264.

Hoare, D., Krause, J., Peuhkuri, N., Godin, J.G., 2000. Body size and shoaling in fish. J.Fish Biol. 57, 1351–1366.

Majkowski, J., 2007. Global Fishery Resources of Tuna and Tuna-like Species. FoodAgriculture Org.

Marra, G., Radice, R., 2013. A penalized likelihood estimation approach to semipara-metric sample selection binary response modeling. Electron. J. Stat. 7, 1432–1455.

Meekan, M., Jarman, S., McLean, C., Schultz, M., 2009. DNA evidence of whale sharks(Rhincodon typus) feeding on red crab (Gecarcoidea natalis) larvae at ChristmasIsland, Australia. Mar. Freshw. Res. 60, 607–609.

Morato, T., Machete, M., Kitchingman, A., Tempera, F., Lai, S., Menezes, G., Pitcher, T.J.,Santos, R.S., 2008a. Abundance and distribution of seamounts in the Azores. Mar.Ecol.-Prog. Ser. 357, 17–21.

Morato, T., Varkey, D.A., Damaso, C., Machete, M., Santos, M., Prieto, R., Santos, R.S.,Pitcher, T.J., 2008b. Evidence of a seamount effect on aggregating visitors. Mar. Ecol.Prog. Ser. 357, 23–32.

Pallarés, P., Petit, C., 1998. Tropical tunas: new sampling and data processing strategy forestimating the composition of catches by species and sizes. Col. Vol. Sci. Pap. ICCAT48, 230–246.

Pham, C., Canha, A., Diogo, H., Pereira, J.G., Prieto, R., Morato, T., 2013. Total marinefisheries catch for the Azores (1950-2010). ICES J. Mar. Sci. 70, 564–577.

Pianet, R., Pallarés, P., Petit, C., 2000. New Sampling and Data Processing Strategy forEstimating the Composition of Catches by Species and Sizes in the European PurseSeine Tropical Tuna Fisheries. IOTC.

Pitcher, T.J., Parrish, J.K., 1993. Functions of shoaling behaviour in teleosts. In: Pitcher,T.J. (Ed.), Behaviour of Teleost Fishes. Chapman and Hall, London, pp. 363–439.

R-Core-Team, 2013. R: A Language and Environment for Statistical Computing.Rohner, C.A., Richardson, A.J., Jaine, F.R., Bennett, M.B., Weeks, S.J., Cliff, G., Robinson,

D.P., Reeve-Arnold, K.E., Pierce, S.J., 2018. Satellite tagging highlights the im-portance of productive Mozambican coastal waters to the ecology and conservationof whale sharks. PeerJ. 6 e4161.

Román, M.H., Aires-da-Silva, A., Vogel, N.W., 2018. Whale Shark Interactions with theTuna Purse-Seine Fishery in the Eastern Pacific Ocena. Summary and analysis ofavailable data. WORKING GROUP ON BYCATCH. La Jolla, California: INTER-AMERICAN TROPICAL TUNA COMMISSION.

Romanov, E.V., 2002. Bycatch in the tuna purse-seine fisheries of the western IndianOcean. Fish. Bull. 100, 90–105.

Rowat, D., Brooks, K.S., 2012. A review of the biology, fisheries and conservation of thewhale shark Rhincodon typus. J. Fish Biol. 80, 1019–1056.

Sampaio, C.L., Leite, L., Reis-Filho, J.A., Loiola, M., Miranda, R.J., José de Anchieta, C.,Macena, B.C., 2018. New insights into whale shark Rhincodon typus diet in Brazil: anobservation of ram filter-feeding on crab larvae and analysis of stomach contentsfrom the first stranding in Bahia state. Environ. Biol. Fishes 1–9.

J. Fontes, et al. Fisheries Research 229 (2020) 105598

8

Santos, R.S., Hawkins, S., Monteiro, L.R., Alves, M., Isidro, E.J., 1995. Marine research,resources and conservation in the Azores. Aquat. Conserv.: Mar. Freshw. Ecosyst. 5,311–354.

Santos, R.S., Porteiro, F.M., Barreiros, J.P., 1997. Marine fishes of the Azores: an anno-tated checklist and bibliography. Arquipelago - Life Mar. Sci. 1–231 Sup 1.

Silva, M.A., Feio, R., Prieto, R., Gonçalves, J., Santos, R., 2002. Interactions between

cetaceans and the tuna fishery in the Azores. Mar. Mamm. Sci. 18, 893–901.Whittingham, M.J., Stephens, P.A., Bradbury, R.B., Freckleton, R.P., 2006. Why do we

still use stepwise modelling in ecology and behaviour? J. Anim. Ecol. 75, 1182–1189.Woods, J.D., Barkmann, W., 1995. Modeling oligotrophic zooplankton production - sea-

sonal oligotrophy off the Azores. ICES J. Mar. Sci. 52, 723–734.

J. Fontes, et al. Fisheries Research 229 (2020) 105598

9