Is Whale Shark Tourism Ecologically Sustainablein Mozambique?

Is whale shark tourism ecologically sustainablein Mozambique?

Running head: Whale shark tourism in Mozambique

Peter J. Haskell¹´², Andrew McGowan¹, Anna Westling³, Adriana Méndez-Jiménez³, Christoph A.

Rohner²´⁴´⁵, Kym Collins3, Marcela Rosero-Caicedo³, Jodi Salmond3, Ara Monadjem⁶, Andrea D.

Marshall²´⁷´⁸, Simon J. Pierce²´³´⁷´⁸*

¹University of Exeter, School of Biosciences, Cornwall Campus, Penryn, Cornwall, TR10 9EZ, UK

²Marine Megafauna Association, Tofo Beach, Inhambane, Mozambique

³All Out Africa Research Unit, Lobamba, Swaziland

⁴Biophysical Oceanography Group, School of Geography Planning and Environmental

Management, The University of Queensland, St Lucia QLD 4072, Australia

⁵Climate Adaptation Flagship, CSIRO Marine and Atmospheric Research, EcoScience Precinct,

Dutton Park QLD 4102, Australia

⁶All Out Africa Research Unit, Department of Biological Sciences, University of Swaziland,

Kwaluseni, Swaziland

⁷ECOCEAN USA, Tofo Beach, Inhambane, Mozambique

⁸Ocean Revolution (Mozambique), Tofo Beach, Inhambane, Mozambique

*Email:

ABSTRACT:

The whale shark (Rhincodon typus) is a popular focal species in the global marine tourism industry. While this business can create an economic incentive for the conservation ofthis threatened species, tourism itself can negatively impact on sharks in some circumstances.We analysed 689 encounters with at least 142 individual sharks over 2.5 years to assess thebehaviour of whale sharks in the presence of swimmers at Tofo Beach in Mozambique. Sharks ranged in size from 3.0 to 9.5 m in total length, with males constituting the majority (74%).Avoidance behaviours were displayed by the sharks during 64.7% of encounters. Encounterduration decreased significantly from 12 minutes 37 seconds to 8 minutes 25 seconds in caseswhere the sharks expressed avoidance behaviours, indicating that tourist interactions did affectthe short-term behaviour of sharks. The presence of injuries or scars on a shark, its feedingactivity and the number of swimmers present all had a significant positive relationship withencounter duration. Mean encounter duration, the overall expression of avoidance behaviour,and the likelihood of an individual shark exhibiting avoidance behaviours did not show asignificant trend over the study period. Although potential tourism impacts are likely to bemitigated by the non-breeding status and transient behaviour of sharks at this aggregation site,the repeated exposure of some sharks to tourists suggests that concerted effort should bemade to implement best-practice standards amongst operators to ensure long-termsustainability.

KEY WORDS: behavioural observations, ecotourism, conservation, monitoring, risk assessment

INTRODUCTION

Viewing sharks in their natural setting is an increasingly popular tourism activity (Smith et al.2010, Gallagher & Hammerschlag 2011) and the income accrued can create a powerfulincentive to manage these charismatic species as a non-consumptive resource (Brunnschweiler2010, Vianna et al. 2010, Clua et al. 2011). In some cases, shark tourism can even provide adirect economic alternative to fishing (Pine et al. 2007, Gallagher & Hammerschlag 2011). Withthis in mind, tourism has been mooted as a potential component of shark conservationstrategies (Topelko & Dearden 2005) and has been a primary consideration in increased legalprotection for elasmobranchs or their habitats in a number of countries (e.g. Topelko &Dearden 2005, Graham 2007, Pine et al. 2007, Anderson et al. 2011). However, recent studieshave documented situations in which marine tourism has negative impacts of its own on focalelasmobranch species in the form of behavioural changes and increased energetic costs (Pierceet al. 2010, Smith et al. 2010, Fitzpatrick et al. 2011). An improved understanding of the actualor potential effects of tourism interactions on focal species is important for mitigating oravoiding longer-term effects, and therefore ultimately safeguarding employment and tourisminfrastructure. This leads us to the main question posed in this paper: is whale shark Rhincodontypus tourism in Mozambique likely to have a detrimental impact on this threatened species?

The whale shark is the world’s largest fish and an iconic tourism species (Gallagher &Hammerschlag 2011). Commercial whale shark interaction tours began at Ningaloo Reef,Western Australia, in 1993 after a seasonal aggregation of the species was discovered within the Ningaloo Marine Park (Davis et al. 1997).Whale sharktourism industry was valued atAU$6.0 million annually in 2006 at this site (Catlin et al. 2010). Whale shark tourism industrieshave now developed at several locations in all three tropical oceans. Most of these destinationsare based on the presence of predictable seasonal feeding aggregations of sharks exploitingephemeral bursts in local productivity such as mass fish, or coral, spawning events (Taylor 1996,Heyman et al. 2001; de la Parra Venegas et al. 2011). Whale shark tourism provides a significantboost to many regional economies, particularly in developing countries where the majority ofestablished destinations are located. The total global revenue from whale shark tourism wasprovisionally estimated to be US$42 million in 2007 (Graham 2007).

Until recently, whale sharks were targeted by fisheries in several countries including thePhilippines (Alava et al. 2002), India (Pravin 2000), the Maldives (Anderson & Waheed, 2001)and Taiwan (Chen & Phipps 2002), which led to declining catch rates and eventual fisheriesclosures. The species was assessed as globally Vulnerable on the IUCN Red List of ThreatenedSpecies (Norman 2005) and is one of the few shark species listed on international conservationconventions such as the Convention on International Trade in Endangered Species (Appendix II)and the Convention on Migratory Species (Appendix II). While the economic value or potentialfor whale shark tourism has helped to justify legal protection for the species in some of thesecountries, such as the Philippines and the Maldives, concerns have also been raised thatspecialist tourism industries could negatively impact the sharks. Sightings of whale sharks atGladden Spit in Belize declined between 1998 and 2003 (Graham & Roberts 2007) and, basedon further anecdotal reports from guides, stayed low up until at least 2007 (Graham 2007). Graham (2007) suggested that the rapid increase in diver numbers 89 at this site may have led todisturbance of snapper spawning behaviour (the main driver of whale shark presence) and thewhale sharks themselves, although a dedicated study on diver disturbance at this site did notfind such an effect (Heyman et al. 2010). Propeller injuries on whale sharks from small boatshave also been noted at several aggregation sites (Cárdenas-Torres et al. 2007, Rowat 2007,Rowat et al. 2007, Speed et al. 2008,). Examination of the short-term behavioural responses ofsharks to tourists and boats at Ningaloo Reef (Norman 1999), Donsol in the Philippines (Quiros2007) and Tofo Beach in Mozambique (Pierce et al. 2010) demonstrated that sharks routinelydisplay avoidance behaviours, including banking, eye-rolling, fast swimming and diving, inresponse to close approaches by swimmers or boats.

The ecological, social and economic sustainability of whale shark tourism at Ningaloo Reef wasreviewed by Mau (2008). Although long-term empirical data on whale shark behaviour was notavailable from the area, the industry was judged to be ecologically sustainable based on thelack of observed interruption of feeding behaviour, which may take place largely at night or atleast outside tourist interaction times (Taylor 2007), the regular re-sightings of philopatricsharks (Holmberg et al. 2008, 2009), and the lack of reproductive behaviour observed in thisjuvenile male-biased population (Meekan et al. 2006, Norman & Stevens 2007). However,concerns were raised about the potential for injury from boat strikes (Speed et al. 2008) and anapparent decline in mean size over time (Bradshaw et al. 2008). Aside from Ningaloo Reef,where the industry has been relatively well-studied, the longer-term sustainability of shark tourism has not been explicitly considered. Studies on marine mammals have concluded that tourism can cause long-term cumulative impacts on the focal 111 species, such as habitat shifts(Bejder et al. 2006b; Allen and Reid 2000), heightened stress (Bejder et al. 1999), and areduction in foraging behaviour (Carrera et al. 2008). Individuals who are forced to divert theirenergies from normal behaviours, such as feeding, to avoidance behaviours may suffer from anoverall reduction in biological fitness (Bejder et al. 1999). Given the threatened status andeconomic value of whale sharks, this topic is evidently worthy of further exploration.

In this study we examine the tourism industry at Tofo Beach (Praia do Tofo) in Mozambique, aninternational hot-spot for whale shark encounters (Pierce et al. 2010). Whale shark interactionsare a key draw for international divers visiting the country, but no official management hasbeen instituted to date raising questions over the sustainability of the industry (Pierce et al.2010, Tibiriçá et al. 2011). A relatively high density of whale sharks are present throughout theyear in conjunction with localised productivity and meso-scale oceanographic features that maytransport whale sharks through the study site (Rohner et al. In press). Previous work at TofoBeach has examined how interactions can be managed so as to minimise the potential forshort-term negative impacts on the sharks (Pierce et al. 2010). Here we extend and expandupon that dataset to evaluate whether longitudinal encounter data reveal evidence of anincreasing frequency of avoidance behaviours overall, and whether individual-based analysesshow evidence of changing avoidance behaviours over time. The implications of these data arethen explored to evaluate the longer-term implications for the ecology of whale sharks at thissite.

MATERIALS AND METHODS

Study site and commercial snorkelling trip procedures

The small village of Tofo Beach (23°51’ S, 35°32’ E) is situated in the Inhambane province ofMozambique, approximately 400 km north-east of the country’s capital city, Maputo (Fig. 1). Afull site overview and description of the commercial whale shark tourism industry in Tofo wasprovided by Pierce et al. (2010). Briefly, operators offer daily two-hour ‘ocean safari’ snorkellingtrips that aim to find and swim with whale sharks and other marine megafauna. Trips generallytake place between 11 am and 2 pm, when high light penetration into the water aids the visualsearch for animals. During each trip, vessels typically survey a 6 km stretch of coastline south ofTofo between the surf line and 1000 m from the shore in waters ranging in depth fromapproximately 5 to 30 m. All trips were conducted from a single tour operator using an 8.2 mrigid-hull inflatable boat. Sharks were located through visual inspection of the water surface,where it was often possible to spot their large, dark silhouettes or exposed fins. The boat wasthen positioned in relation to the shark’s direction of travel and clients entered the water tointeract with the shark.

Data collection

Data were collected from January 2008 until June 2010, except July and August 2008 when nosampling trips took place. Upon locating a whale shark, observers entered the water alongside clients and recorded pre-determined observations whilst snorkelling 155 with the shark. The totalnumber of swimmers and environmental characteristics, including weather conditions(categorised as sunny, slightly overcast, overcast, or raining), Beaufort sea state andunderwater visibility were recorded. The total length (TL) of the shark was estimated visually(Rohner et al. 2011) and the sex identified by the presence or absence of claspers on the pelvicfins. The presence and location of any injuries or scars were noted and categorised post-hoc aseither ‘major’ or ‘minor’ following the definitions of Speed et al. (2008). A basic ethogram ofeach shark’s behaviour in the presence of swimmers was also produced to record the followingcharacteristics: slow swimming (SS), equating to normal behaviour; fast swimming (FS), wherethere was an obvious increase in the shark’s tail-beat frequency; diving (D), where the sharkdived away from the surface; banking (B), where the shark rolled its back toward swimmer/s;changing direction (CD), where the shark altered its direction of swimming in the presence ofswimmers; and any other obvious avoidance behaviours (A), such as the violent shudderreported by Quiros (2007). Observations of feeding behaviour were also noted when theyoccurred. The total encounter duration, defined as the time between the first swimmerentering the water and the last swimmer returning to the boat, was recorded (in minutes)following each interaction. When possible, standardised identification (ID) photographs of boththe left and right flanks were taken for each shark and uploaded to the global ECOCEAN whaleshark photo-identification library (Arzoumaniam et al. 2005). Each of these encounters werethen assigned to a new or previously-identified shark in the library.

Defining avoidance behaviours

A lack of opportunity to observe the animal under undisturbed conditions is a commondifficulty when evaluating the impact of human presence on animal behaviour (Bejder andSamuels 2003; Quiros 2007). It was difficult to establish whether behaviours that wereexhibited by whale sharks in the presence of swimmers would also occur in a natural,undisturbed setting. Within-effect comparisons (i.e. behaviour in the presence of swimmers)were therefore employed instead of attempting to determine ultimate cause and effectrelationships. This approach is useful in cases where there is no baseline or control data, andfocuses on assessing behavioural responses under gradations of the effect (Bejder and Samuels2003). Avoidance responses are not mutually exclusive and lie along a gradient ranging from noresponse to a high level of response in cases where multiple avoidance behaviours areexhibited (Quiros 2007). Behaviours that are typically classified as avoidance behaviours inwhale sharks include FS, D, B and CD (Norman 1999, Quiros 2007, Pierce et al. 2010). Whenthese or A were observed during an interaction they were each assigned a score of one. Thetotal score at the end of each encounter was thus equal to the total number of avoidancebehaviours observed, which was then used as the overall avoidance score during analysis. Allstatistical analyses were carried out using R version 2.11.0 ( with additionalsoftware packages “MASS” and “nlme”.

Analysis of short-term responses

The relationship between avoidance score and encounter 198 duration was tested using aGeneralised Linear Mixed Model (GLMM) approach. Encounter duration, log transformed toensure that error structure was normally-distributed, was used as the response variable.Avoidance score was used as the explanatory variable and shark ID was added as a randomeffect to avoid pseudo-replication. Due to the significant relationship between avoidance scoreand encounter duration (see results, and also Pierce et al. 2010), encounter duration was usedin all further analyses as a proxy measure for avoidance. A further GLMM was fitted todetermine which variables significantly affected encounter duration. Encounter duration wasentered as the response variable and tested against nine main effect explanatory variables:whale shark variables (size, sex, presence/absence of scarring or feeding behaviour,respectively), environmental variables (weather [clear, partially overcast, overcast or rain], seastate (0-4) and underwater visibility), number of swimmers present during an encounter andthe Julian day (Day 1 = 9th January 2008). Second order interactions were tested between allcombinations of whale shark variables, all combinations of environmental variables, size, sexand scarring with all environmental variables and between the number of swimmers andunderwater visibility. Year and month of the encounter were entered as nested random effects.Only encounters in which shark ID had been confirmed were used in the analysis; a singleinteraction was randomly selected from those individuals for which multiple encounters hadbeen recorded. A generalised linear model (GLM) was used to test for a significant relationshipbetween the level of scarring on an individual and its total length.

Analysis of longitudinal responses

A GLM was fitted using the number of encounters as the response variable and search effort (inhours) and year as explanatory variables to establish whether encounter rate changed over thecourse of the study. As the dataset ended in June 2010, a second GLM was fitted to six-monthlyperiods using the number of encounters as the response variable and search effort and sixmonth block as explanatory variables. A quasipoisson error structure was used for bothanalyses due to overdispersion of residual deviance. A third GLM was used to test for a changein encounter duration during the study period using log-transformed encounter duration as theresponse variable and month and year as explanatory variables. Chi-square was used to test fora population-level increase in avoidance response over the five six-month periods. Ageneralised LME model was used to test whether the likelihood of avoidance was related to thenumber of previous encounters in which the whale shark had been successfully identified andits behaviours recorded over the study period. The ECOCEAN library was used to establish howmany previous encounters had been recorded for each individual shark. A binomial score foravoidance was used as the response variable, encounter number as the explanatory variableand shark ID as a random factor. Significance was accepted at the 95% confidence interval in allanalyses. The standard deviation for all appropriate data values is presented.

RESULTS

Encounter data

A total of 689 whale shark encounters were recorded from 328 trips between January 2008 andJune 2010, with a total search effort of 476.5 hrs. Between 0 and 14 encounters were recorded

on each trip, with an overall mean of 2.73 ± 0.17 trip-1, equating to 1.46 ± 0.10 sharks hr-1 ofsearch effort. At least one shark was spotted on 77.1% of trips. There was no significant trend insightings over the course of the study (Linear Regression, 27 d.f., P = 0.260) (Fig. 2). A total of142 individual sharks were positively identified using the ECOCEAN photo-identification library.Of 128 sexed sharks, 33 were female (25.8%) and 95 were male (74.2%) which was significantlydifferent from a 1:1 sex ratio (Chi-square, χ1 249 = 30.031, P = <0.001). Size ranges were 3.0 to 8.5 mTL for female sharks (mean 5.85 ± 1.28 m) and 3.0 to 9.5 m TL in male sharks (mean 6.14 ± 1.24m) (Fig. 3), with no significant difference between the sexes (t-test, t = -1.144, df = 125, P =0.255). The total number of encounters (counted as the number of days in which interactionswith that shark were recorded) with identified sharks ranged from one to five over the courseof the study, although the inclusion of additional data from the ECOCEAN library showed thatthe individual whale sharks considered in this study had been exposed to swimmers up to 12times before June 31st 2010, with a mean of 3.08 ± 2.13 encounters (Fig. 4). The reliance onsuitable photographs having been submitted to the library means that these figures representthe minimum number of previous encounters. Scarring was observed on 53.3% of identifiedindividuals and whale sharks were observed feeding during 19.5% of all encounters.

Short-term encounter duration and avoidance responses

Overall mean encounter duration was 9 minutes 46 seconds ± 263 8 minutes 18 seconds (N = 613).One or more avoidance responses were recorded during 67.5% of encounters. A total of 184encounters, from 132 identified sharks, were evaluated to determine that encounter durationwas significantly related to expressed avoidance response (GLMM, χ23 = 15.046, P = 0.002). Themean duration of an encounter when a shark showed no avoidance was 12 minutes 37 seconds± 8 minute 58 seconds (N = 190), significantly longer than encounters with sharks that exhibitedavoidance (GLMM, χ21 = 14.255, P < 0.001). Durations did not differ significantly amongencounters where one or more avoidance behaviours were expressed (GLMM, χ22 = 0.791, P =0.673), which had a mean encounter duration of 8 minutes 25 seconds ± 7 minutes 21 seconds(N = 397).