Biological Conservation 158 (2013) 280–286

Contents lists available at SciVerse ScienceDirect

Biological Conservation

journal homepage: www.elsevier .com/locate /b iocon

Effect of illegal harvest on apparent survival of Amazon Riverdolphins (Inia geoffrensis)

Vanessa J. Mintzer a,b,⇑, Anthony R. Martin c, Vera M.F. da Silva d, Andrew B. Barbour b, Kai Lorenzen b,Thomas K. Frazer a,b

a School of Natural Resources and Environment, University of Florida, 103 Black Hall, Box 116455, Gainesville, FL 32611, USAb Fisheries and Aquatic Sciences Program, School of Forest Resources and Conservation, University of Florida, 7922 NW 71st St., Gainesville, FL 32653, USAc Centre for Remote Environments, University of Dundee, 23, Springfield, Dundee DD1 4JE, UKd Instituto Nacional de Pesquisas da Amazônia – INPA/Laboratório de Mamíferos Aquáticos, Av. André Araújo 2936, Manaus, Amazonas 69060-001, Brazil

a r t i c l e i n f o a b s t r a c t

Article history:Received 30 June 2012Received in revised form 5 October 2012Accepted 11 October 2012Available online 29 November 2012

Keywords:Barker modelBotoCalophysus macropterusMark-recaptureBaitPoaching

0006-3207/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.biocon.2012.10.006

⇑ Corresponding author at: School of Natural RUniversity of Florida, 103 Black Hall, Box 116455, Ga+1 352 359 5633; fax: +1 352 392 9748.

E-mail addresses: vjs@ufl.edu (V.J. Mintzer), botucuxi@inpa.gov.br (V.M.F. da Silva), snook@ufl.edu (edu (K. Lorenzen), frazer@ufl.edu (T.K. Frazer).

The Amazon River dolphin (Inia geoffrensis), or boto, is illegally harvested for use as bait in fisheries for thecatfish Calophysus macropterus. To determine the effect of this harvest, we estimated apparent survival fora boto population in the central Brazilian Amazon where direct harvest is known to have occurred since2000. For our analysis, we used capture and recapture/resighting data of 528 marked botos over a 17-yearperiod (1994–2011). Time-dependent models estimated that apparent survival after the first reports ofharvest (/ = 0.899; SE = 0.007) was significantly lower than in years prior to harvest (/ = 0.968;SE = 0.009). The decline in apparent survival suggests that current harvest rates exceed conservation lim-its and may be unsustainable. This issue requires the attention of natural resource managers from allcountries of the Amazon basin, as the harvest is widespread and decline in survival could be mirroredin numerous locales.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Interactions with fisheries, mostly through incidental capture infishing gear but also through targeted harvesting are recognized asa key threat to aquatic mammal populations (Clapham and VanWaerebeek, 2007; Costello and Baker, 2011; Jefferson and Curry,1994; Mangel et al. 2010; Northridge and Hofman, 1999; Perrinet al., 1994; Read, 2008; Read et al., 2006; Robards and Reeves,2011; Vidal, 1993). Perhaps the least well quantified aspect ofaquatic mammal–fisheries interactions is the harvesting of mam-mals for use as bait, a practice affecting at least 38 species of aqua-tic mammals worldwide (Diniz, 2011). Direct harvest of aquaticmammals for this purpose is considered an illegal activity in manynations where it takes place; hence, the prevalence of this illicitpractice and subsequent impact has proved challenging to quan-tify. Most studies on the issue only provide information on wherethe illicit harvest occurred without related data on intensity, dura-

ll rights reserved.

esources and Environment,inesville, FL 32611, USA. Tel.:

to@live.co.uk (A.R. Martin),A.B. Barbour), klorenzen@ufl.

tion, or effects of the harvest (Diniz, 2011). However, in a limitednumber of cases where such ancillary data has been provided, find-ings suggest illegal harvests for bait may have considerable im-pacts on targeted populations (e.g. da Silva et al., 2011; Manzurand Canto, 1997; Sinha, 2002).

The Amazon River dolphin (Inia geoffrensis de Blainville, 1817), orboto, has been harvested for use as bait in the fisheries for the catfishknown as piracatinga, mota, simi, zamurito or mapurite (Calophysusmacropterus Lichtenstein, 1819) (da Silva et al., 2011; Gómez et al.,2008; Gómez-Salazár et al., 2012; Loch et al., 2009; PortocarreroAya et al., 2010a; Shostell and Ruiz-Garcia, 2010; Trujillo et al.,2010b, 2010c). During the last decade, demand for C. macropterushas grown in Colombia as other species of catfish once popularlyconsumed in the country have been overfished (Gómez et al.,2008; Petrere et al., 2004; Trujillo et al., 2010b). In fact, demand issuch that an international market has emerged with Brazil, Venezu-ela, Peru, and Bolivia, all contributing to the export of this food fishto Colombia, where it is often sold under the name capaz, one of thedepleted catfish species from the Magdalena River (Diniz, 2011;Gómez et al., 2008; Trujillo et al., 2010b, 2007). More recently, con-sumption of C. macropterus has also gained popularity in a few Bra-zilian cities where it is marketed as douradinha, a ‘‘new’’ species offood fish (da Silva et al., 2011; Trujillo et al., 2010b). Thus, althoughthe boto is considered the least endangered of the river dolphins

V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286 281

(Inioidea and Platanistidae; Leatherwood and Reeves, 1994; Smithand Smith, 1998), the demand for C. macropterus and related harvestof boto for bait raises substantial concern.

At present, harvest of botos for bait is known to occur in at leasttwelve locations in four of the five Amazonian countries with C.macropterus fisheries (Diniz, 2011), including in and around theMamirauá Sustainable Development Reserve (MSDR) upriver ofthe Brazilian town of Tefé. Although the harvest in this regionmay have started as early as the mid-1990s, the first cases in MSDRwere documented in 2000 (da Silveira and Viana, 2003; Estupiñánet al., 2003). Using C. macropterus landings data from Tefé, da Silvaet al. (2011) estimated that roughly 1650 botos are killed in thearea per fishery season. Moreover, the same authors reported a10% average annual decline in boto abundance in one of theMSDR’s main channels beginning in 2001, a trend coincident withthe first documented occurrence of the illegal harvest. These find-ings highlight the need to further explore the impact of harvestingfor bait on the boto population.

Herein, we provide estimates of apparent survival for the botopopulation occurring in and around the Mamirauá Lake System(MLS) at the southern edge of the MSDR. Using mark-recapture/resighting models, we explore changes in survival between periodsof varying harvest pressure. We expected that survival estimatesprior to the documented start of the harvest, i.e. 1994–2000, wouldbe greater than estimates for the more recent period for which datawere available, i.e. 2000–2011. Additionally, we compared survivalbetween the sexes, with the expectation that survival of maleswould be significantly lower than that of females, based on differ-ences in morphology and behavior. Males are larger (55% heavierand 16% longer than females), exhibit high levels of intermaleaggression (Martin and da Silva, 2006), and show preference forriver habitat outside the protected area boundary (Martin and daSilva, 2004b).

2. Methods

2.1. Study area

All data were collected in and around the MLS located at theconfluence of the Solimões and Japurá rivers, 20 km upriver of

Fig. 1. Map of study site, the Mamirauá Lake System and surrounding areas, locatedMamirauá Sustainable Development Reserve in Amazonas State, Brazil (GIS layers: IUCN

Tefé in Amazonas State, Brazil (Fig. 1). The study area straddlesthe southern edge of the MSDR. Typical of whitewater floodplainor várzea habitat, the MLS has numerous channels and lakes anda diverse fish fauna that varies seasonally in response to waterfluctuations. During May and June, water levels rise 11–15 m andthe MLS is completely flooded. Lowest water levels typically occurbetween September and November. These water fluctuationsdetermine what habitats are available for the boto population.While botos may be restricted to the deepest channels and lakesduring the low water months, they may swim freely and enter sub-merged forest at high water. Previous estimates suggest approxi-mately 260 botos occur in or near the 225 km2 MLS year-round(Martin and da Silva, 2004a, 2004b). Half of these individuals exhi-bit strong site fidelity to the study area, while others are consid-ered transient and visit for shorter periods (Martin and da Silva,2004a).

2.2. Capture and recapture/resight

Our capture–recapture and resight data spans 17 years(January1994–November 2011). These data were collected throughProjeto Boto, a river dolphin research program in the MSDR.Capture–recapture and marking of botos occurred approximately3 weeks each year. With few exceptions, we captured botos withseine nets during low water at the entrance of the MLS. When cap-tured, we freeze-branded botos with a unique code in areas allow-ing for maximum contrast and visibility (usually the dorsal fin andflank). When previously marked botos were captured, we recordedtheir code and rebranded the individual if necessary. Furtherdescription of capture procedures is available in da Silva and Mar-tin (2000) and Martin et al. (2006). These data were collected withthe approval of the Instituto Chico Mendes de Conservação da Bio-diversidade (Sistema de Autorização e Informação em Biodiversid-ade #13462-1).

In addition to the capture–recapture events, we conductedyear-round observational work to provide resightings betweencapture events. We conducted sighting surveys throughout theMLS and surrounding areas, including segments of the main rivers.Due to varying water levels, we did not evenly distribute effortacross the entire study area. Most areas, however, were surveyed

at the junction of the Japurá and Solimões rivers in the southern segment of theand UNEP 2010; DCW and GADM downloaded from www.diva-gis.org).

Year

Hou

rs o

f obs

erva

tiona

l effo

rt

1000

1100

1200

1300

1400

1500

1600

1700

2000 2002 2004 2006 2008

Fig. 2. Hours of observation work conducted in and around the Mamirauá LakeSystem to provide resight data of Inia geoffrensis between primary capture events.Effort from 1999 to 2009 is displayed showing an increasing trend over time. Thesesighting surveys were typically conducted by 2–3 observers from a 4 m aluminumskiff, powered by a 15HP outboard motor.

282 V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286

at least once per week by 2–3 observers from a 4 m skiff. We in-cluded additional sightings from other platforms, such as the Pro-jeto Boto floating laboratory. When a marked boto was sighted, werecorded its unique code and our percent confidence in the identi-fication (ranging from 50% to 100%, but normally 100%). Since1999, an average of almost 1500 h of observational work has beenconducted each year with an increasing trend over time (Fig. 2).

2.3. Survival model selection and data analysis

We estimated apparent survival of the marked population withthe Barker model (Barker, 1997, 1999) in Program MARK (Whiteand Burnham, 1999). This model allows for physical capture–re-capture during discrete primary periods, and continuous resightingdata during secondary periods (the intervals between primaryperiods). We created capture histories for marked individuals cap-tured between 1994 and 2011, using resightings made with 100%confidence. We coded for uneven time intervals by scaling365 days to equal an interval of length 1.

Program MARK estimates seven parameters for the Barker mod-el (Table 1) (White and Burnham, 1999). As resightings did not oc-cur over the entire geographic range of the study population; weestimated apparent survival (/), which includes the effects of emi-gration (/ = true survival x (1 - probability of emigration)). Wetransformed real parameter values [0,1] to b-values [�1,1] withthe sin-link or logit-link function for numerical optimization. Weused simulated annealing for optimization due to a prior simula-tion that identified problematic likelihood surfaces leading to con-vergence failures when using the Newton-Raphson method withthe Barker model.

Table 1Barker joint data model parameter definitions.

Parameter Definition

/i The probability that an animal remains alive and available for recapturepi The probability an animal at risk of capture at i is captured at iri The probability an animal that dies in i, i + 1 is found deadRi The probability an animal that survives from i to i + 1 is resighted aliveR0i The probability an animal that dies in i, i + 1 without being found deadFi The probability an animal at risk of capture at i is at risk of capture at iF 0i The probability an animal not at risk of capture at i is at risk of capture

differs from the one provided in Barker (1997) in order to enforce intern

The Barker model is heavily parameterized; hence, we made apriori assumptions on how to treat p, r, and R (Table 1). We fixedr as a time-independent (.) parameter because we had few deadrecoveries (only 12 marked carcasses were found during the studyperiod) and therefore lacked sufficient data to inform time-depen-dent estimates. We defined R and R0 as time-dependent parametersto allow for changes in observation effort through time (Fig. 2).Both of these parameters were allowed to vary between 3-yearintervals (3y) to reduce the parameter count. Capture probability(p) was treated as a time-dependent parameter (t) because we ex-pected p to vary according to the environmental conditions of eachcapture expedition. We created a series of models that allowed /, F,and F0 (Table 1) to vary with time dependence (t) or independence(.). We also built models that allowed / and F to vary according tosex (G). Additionally, because the illegal harvest of botos in theMSDR started approximately in 2000, we created models where/ was allowed to vary between two periods: 1994–2000 (i.e. pre-harvest) and 2000–2011 (i.e. harvest). The data were partitionedon November 20, 2000, the last day of the 2000 capture expedition.

We conducted a median c goodness-of-fit test on the globalmodel /(G � t)p(G � t)r(G � t)R(G � t)R0(G � t)F(G � t)F0(G � t) to as-sess overdispersion (White and Burnham, 1999). We then appliedthe estimated c (variance inflation factor) to the model set. Valuesthat do not fall within 1 6 c 6 4 indicate a structural lack of fit(Burnham and Anderson, 2002). We used Akaike’s Information Cri-terion (AIC) values to rank models (Akaike, 1973). Lower AIC valuesindicate models that better explain the variation in the data withmaximum parsimony (Taper et al., 2008). Because we applied theestimated c to our model set, we used the small-sample, c-cor-rected version of AIC, QAICc, for model ranking. Models with QAICc

values that differed by less than 2 were considered equal(Burnham and Anderson, 2002, 2004). We tested for a significantdifference between apparent survival estimates corresponding todifferent time periods and sex by examining the 95% confidenceintervals (CIs) of the / b-values representing the time or sex effect(if the 95% CIs of the b-values did not include 0.0, a significant ef-fect was determined at the a = 0.05 level).

3. Results

We uniquely branded 528 botos (256 females and 272 males).Excluding botos first marked in 2011, we resighted 87.71% of fe-males and 88.93% of males at least once. We recaptured an averageof 25.56 (±17.90 SD) branded individuals each year (Fig. 3).

The median c goodness-of-fit test resulted in c ¼ 1:162, wellwithin the acceptable range, and we adjusted all model resultswith this value. QAICc strongly supported a model with fullytime-dependent survival estimates (Table 2; Model 1). This modelestimated a significant difference in F by sex, with estimates of /= 0.808 (SE = 0.051) for females and / = 0.706 (SE = 0.064) formales (Table 2; Model 1). Capture probability (p) varied consider-ably between years, with an average of p = 0.253. This model sug-gested declining apparent survival through time, with an average

from i to i + 1

some time between i and i + 1is resighted alive in i, i + 1 before it died+ 1

at i + 1 (this definition, as applied in Program MARK White and Burnham (1999),al constraints)

0

50

100

150

200

250

1995 2000 2005 2010Year

Cou

nt

SexF

M

TypeRecapture

Resighting

Fig. 3. Capture, recapture, and resight counts from 1994 to 2011 of Inia geoffrensisoccurring in and around the Mamirauá Lake System. Capture counts (solid lines)displayed by sex represent the cumulative number of individuals marked. Approx-imately 3 weeks each year were dedicated to the capture–recapture and marking ofbotos. Observational work was conducted between primary capture events toobtain resighting data.

Table 2QAICc table from Barker survival model results.

Model QAICc QAICc QAICc weight k

1. /(t)F(G)F0(t) 6097.76 0.00 0.71 692. /(t)F(.)F0(t) 6099.87 2.11 0.25 683. /(t)F(t)F0(t) 6105.55 7.79 0.01 844. /(t)F(t)F0(.) 6105.57 7.80 0.01 695. /(94-00)(00-11)F(G)F0(t) 6106.82 9.06 0.01 526. /(G�(94-00)(00-11)F(G)F0(t) 6109.41 11.65 0.00 547. /(.)F(G)F0(t) 6128.11 30.36 0.00 518. /(G)F(G)F0(t) 6128.53 30.78 0.00 52

The parameter of primary interest was apparent survival (/). Time-dependence (t)and sex effect (G) were represented with the associated symbols. Parameters p, r, R,and R0 were fixed through a priori assumptions (see Section 2.3). Number of esti-mated parameters (k) is listed for each model.

0.85

0.90

0.95

1.00

1994 - 2000 2000 - 2011

Time interval

Appa

rent

sur

viva

l (ye

ar1 )

Fig. 5. Apparent survival estimates for the pre-harvest (January 1994–November2000) and harvest period (November 2000–2011) for Inia geoffrensis occurring inand around the Mamirauá Lake System. These estimates were the results of themodel: /(94-00)(00-11)p(t)r(.)R(3y)R0(3y)(F(G)F0(t). The difference in apparentsurvival probability between periods was found to be significant at the a = 0.05level.

V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286 283

annual apparent survival of 0.912 over the study’s duration (Fig. 4).When estimating apparent survival separately for the pre-harvest

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1995 2000 2005 2010Year

Appa

rent

Sur

viva

l (y e

a r1 )

Fig. 4. Apparent survival estimates for 1994–2011 for Inia geoffrensis occurring inand around the Mamirauá Lake System. Apparent survival estimates displayed werethe results of the Barker model where / was treated as a fully time-dependentparameter: /(t)p(t)r(.)R(3y)R0(3y)(F(G)F0(t).

and harvest periods, we found a significant difference (pre-harvest:/ = 0.968, SE = 0.009; harvest period: / = 0.899, SE = 0.007) (Ta-ble 2; Model 5; Fig. 5).

The model including sex as a group attribute of survival (Ta-ble 2; Model 8) estimated a non-significant difference in annualapparent survival between females (/ = 0.919, SE = .008) and males(/ = 0.904, SE = 0.009). Moreover, no significant difference wasfound between apparent survival of females and males for thepre-harvest (females: / = 0.973, SE = 0.011; males: / = 0.963,SE = 0.014) or harvest period (females: / = 0.906, SE = 0.009;males: / = 0.891, SE = 0.010) (Table 2; Model 6).

4. Discussion

4.1. Model performance

The Barker model fit the data well ðc ¼ 1:162Þ and accounted forvariable sighting effort with time-dependent estimates of the res-ighting parameter, R. Although botos are known to occasionallyemigrate from the study area for several years, it is unlikely thistemporary emigration induced a major negative bias in apparentsurvival estimates as the Barker model is robust to emigrationdue to its parameterizations (Barker, 1997; Horton and Letcher,2008). Additionally, the c value near 1 suggested a lack of majormodel assumption violations.

4.2. Survival estimates

Our results indicate that annual apparent survival significantlydecreased by 0.069 after the documented start of the harvest.The apparent survival estimate of 0.968 for the pre-harvest periodis consistent with boto life history, which is characterized by traitsassociated with high annual survival such as slow maturation andbirth intervals of approximately 3 years (see da Silva, 2008). Thisapparent survival estimate is comparable to survival estimates ofnon-harvested populations of Tursiops truncatus (Speakman et al.,2010), Tursiops aduncus (Mansur et al., 2012), and Orcinus orca(Matkin et al., 2012) which range from 0.951 to 0.990. However,annual apparent survival estimates for the full study duration (/

284 V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286

= 0.912) and for the harvest period (/ = 0.899; SE = 0.007) are low-er than most published survival estimates of other odontocetes,suggesting a negative effect of harvest on the population.

We did not expect equal apparent survival between the sexes.The species is sexually dimorphic, and males exhibit correspondinghigh levels of intermale aggression that may lead to life threaten-ing injuries (Martin and da Silva, 2006). Furthermore, when a widerange of habitats is available during high water, females show pref-erence for várzea habitats within the MLS, while males are foundmore frequently in the main rivers outside of the protected areaboundary (Martin and da Silva, 2004b). Thus, in theory, males areat a higher risk of being harvested because they spend more timeoutside protected waters. For these two reasons, we expected low-er apparent survival in males. It is possible the harvest is exertingselective pressure on females (if, for example, females are easier tocatch), and increasing mortality of females to the extent that sur-vival of females and males have equalized. However, our resultsdid not show a significant difference between male and femaleapparent survival during the pre-harvest period, suggesting thatsimilar survival probability between the sexes is a naturalcharacteristic.

4.3. Significance of the harvest and conservation implications

Assuming declines in apparent survival were due to lethal asopposed to sub-lethal effects (e.g. changes in behavior), the resultsindicate that mortality of the study population more than doubledbetween the two time periods. Dividing the harvest period appar-ent survival estimate by the pre-harvest period estimate, we calcu-late a reduction in survival corresponding to an annual harvest rate(h) of 0.071. The Potential Biological Removal (PBR) method, com-monly applied to cetaceans, calculates the level of mortality thatwill inhibit a stock from reaching or maintaining its optimum sus-tainable population (Wade, 1998). According to the PBR approach,the maximum allowable harvest rate (hmax) is equal to 1/2Rmax (seeDillingham and Fletcher, 2008), where Rmax is defined as the max-imum annual recruitment rate (Wade, 1998). If we apply an Rmax of0.04, the default value used for cetaceans (Wade, 1998), the hmax

would equal 0.02, a value about three times smaller than our esti-mated h. If we apply a higher Rmax value of 0.06 (theoretically pos-sible in odontocetes with high survival, Reilly and Barlow, 1986;Wade, 1998), the estimated h is still more than double the hmax

of 0.03 calculated by the PBR rule. This comparison suggests thecurrent harvest rates exceed conservation limits commonly ap-plied to cetaceans and could lead to depletion of the population(Dillingham and Fletcher, 2008).

It is possible that factors other than illegal harvest could ex-plain, in part, declines in boto survival. Throughout its geographicrange, dam construction, pollution, entanglement in fishing gear,habitat degradation, boat traffic, and depleted prey resources havebeen identified as important anthropogenic threats to the species(Best and da Silva, 1993; 1989; da Silva and Best, 1996; Gómez-Salazár et al., 2012; Leatherwood and Reeves, 1994; Martin et al.,2004; McGuire and Aliaga-Rossel, 2010; Portocarrero Aya et al.,2010a; Reeves and Leatherwood, 1994; Rosas and Lehti, 1996;Smith and Smith, 1998; Trujillo et al., 2010b, 2010c; Utreraset al., 2010; Vidal, 1993). However, due to the location of the study,we suggest that aside from illegal harvest, only entanglement infishing gear is likely to substantially affect mortality of the studypopulation.

Although incidental mortality in fishing gear has not been stud-ied in most areas in the Amazon, it is known that seine nets andgillnets pose a significant threat to botos (e.g. da Silva and Best,1996; Leatherwood and Reeves, 1994). In the central Amazon,the lampara seine is the most lethal type of net for botos and ac-counts for over 80% of deaths caused by entanglement (da Silva

and Best, 1996). Many of the boto carcasses recovered by ProjetoBoto during the study period showed evidence of entanglement(da Silva and Martin, 2010; Martin et al., 2004). Furthermore,observations suggest fishers kill botos intentionally because theyregard them as competitors, and because botos cause damage tofishing nets (da Silva and Best, 1996; da Silva and Martin, 2010;Loch et al., 2009). The number of artisanal and commercial fleetsin the study region has remained relatively stable in the last twodecades after experiencing a period of fast growth in the 1970sand 1980s (Almeida et al., 2003; Barthem, 1995; Best and da Silva,1989). Thus, although fishery interactions other than boto harvestfor bait are likely causes of additive mortality for our study popu-lation (Martin et al., 2004), there is no indication that such interac-tions have increased considerably since the onset of this study, andare therefore not likely to explain the decrease in apparent sur-vival. However, this threat should be carefully considered in con-servation and management plans, as it was the primary cause ofthe functional extinction of the Yangtze River dolphin (Turveyet al., 2007) and changes in the size, gear, or techniques of Amazo-nian fishing fleets could lead to considerable increases in botomortality.

The significant decline in apparent survival estimated in thisstudy suggests a need for conservation measures aimed at decreas-ing the boto harvest. Currently, two main tools are in place thatshould aid in the protection of the study population. First, the botois protected by Brazilian federal laws, predominantly Law 7.643(1987) that makes it illegal to kill or harass cetaceans in jurisdic-tional waters. However, enforcement of natural resource protec-tion laws in the Amazon is challenging, and efforts are oftencompromised as a consequence of budget constraints, understaff-ing, and other institutional deficiencies (McGuire and Aliaga-Rossel, 2010; Peres and Terborgh, 1995; Trujillo et al., 2010c;Utreras et al., 2010). This appears to be the case in our study region,where only four federal officials have been available to enforceconservation related regulations in an area that exceeds251,000 km2 (Peres and Lake, 2003). Furthermore, lack of aware-ness by fishers of existing legislation has been noted as an imped-iment to the conservation of botos (e.g. McGuire and Aliaga-Rossel,2010; da Silva and Martin, 2010).

The second major conservation mechanism is the partial spatialprotection provided by the MSDR. Although MSDR was not createdwith the specific purpose of protecting river dolphins, the reserveincludes important river dolphin habitat. To our knowledge, theharvest does not occur in the section of our study area that fallswithin the MSDR (see Fig. 1). Our results indicate that both females(F = 0.808; SE = 0.051) and males (F = 0.706; SE = 0.064) exhibithigh fidelity to our study area; therefore, the protected area shouldfunction by protecting the population at least part of the time (pri-marily during rising water when botos enter the várzea, Martin andda Silva, 2004b). If the MSDR did not exist, it is likely the mortalityof our study population would have been higher in the harvest per-iod. Nevertheless, our results imply that the spatial protection inour study area, with its current design and state, may not be suffi-cient in and of itself to protect this population.

Marine protected areas have been shown recently to be effectivein increasing survival rates of small cetaceans by reducing fishery–dolphin interactions (Gormley et al., 2012). These findings suggestthat properly designed protected areas in the Amazon basin mighttranslate into similar benefits for river dolphins. As proposed by theSouth American River Dolphin Protected Area Network (SARDPAN),protected areas created or improved explicitly for river dolphinconservation could have the potential to reduce the impact of fish-ery–river dolphin interactions, as well as help conserve vulnerablefreshwater habitats largely underrepresented in South Americanprotected areas (Portocarrero Aya et al., 2010b). Although much re-mains to be understood about river dolphin ecology, several studies

V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286 285

have provided valuable information on boto habitat preference andmovement that could prove useful in designing new protectedareas or improving existing ones (e.g. Gómez-Salazár et al., 2012;Leatherwood et al., 2000; Martin and da Silva, 2004a, 2004b; Martinet al., 2004; McGuire and Henningsen, 2007; McGuire andWinemiller, 1998). In our study site, for example, extending spatialprotection to adjacent areas where botos aggregate (e.g. mouths ofchannels that connect to the main rivers), could benefit the studypopulation. The effectiveness of new or enhanced conservationmeasures can be monitored and evaluated using the baseline sur-vival estimates provided in this study.

Although improvements made to enforcement and spatial pro-tection could prove to be beneficial, numerous challenges will haveto be overcome to decrease boto harvest. With administrativeboundaries being crossed, the remoteness of the harvest sites,and high economic incentives, both local and international actionand cooperation will be required. Broad scale initiatives like theSARDPAN could facilitate the necessary international collaborationand the implementation of new and improved conservation mech-anisms. Moreover, it is imperative that all drivers of the harvest beconsidered when formulating solutions, including the troublingstatus of the preferred food catfish species in Colombia, as wellas the socio-economic conditions of fishers at the harvest sites.With this in mind, solutions of varying scope and scale should besimultaneously explored. For example, while a widespreadpublic education campaign to halt the trade and consumption ofC. macropterus may be the focus of conservation efforts in Colombia(Trujillo et al., 2010c), more localized initiatives that educatefishers on existing legislation and alternative activities may provesuccessful at harvest sites (da Silva and Martin, 2010; Trujillo et al.2010a).

Interviews with local fishers have revealed the potential eco-nomic importance of the C. macropterus fishery for families in thecentral Brazilian Amazon (da Silva et al., 2011). A skilled fishercan expect to catch anywhere from USD 500 to 1000 worth of C.macropterus with just one boto carcass in one night (da Silvaet al., 2011), a substantial profit considering the low average an-nual income of people in the region. The use of alternative baits(Gómez et al., 2008; Trujillo et al., 2010a, 2010b, 2010c) may bean economically viable solution and encourage a shift away fromthe practice of harvesting botos. In the Ganges River system, forexample, fish oil is an efficient and inexpensive fish attractant sub-stitute for river dolphin (Platanista gangetica) oil (Sinha 2002). Fur-thermore, the role of river dolphins in ecotourism is beingincreasingly recognized throughout the Amazon (PortocarreroAya et al., 2010b; Trujillo et al., 2010c), and in some locales, care-fully regulated tourism activities could form an important link be-tween boto conservation and economic development (Trujilloet al., 2010a, 2010c). Regardless of the corrective actions pursuedto decrease the harvest, providing viable economic alternativesfor fishers is essential to ensure a smooth and more permanenttransition away from the practice of harvesting botos.

4.4. Conclusion

The significant decrease in apparent survival of the study pop-ulation is a serious concern and warrants the attention of naturalresource managers. The issue likely extends beyond the study areaas harvest of botos is known to occur throughout the Amazon ba-sin, and the decline in boto survival could be mirrored in other lo-cales. In areas without designated protection status, harvest levelsare likely to be higher and survival lower than estimated in thisstudy. A timely and transdisciplinary approach is needed to ad-dress all components of the intricate socio-ecological system thatsurround this illegal harvest.

Acknowledgements

This study was part of Projeto Boto, a cooperative agreementbetween the National Amazon Research Institute-INPA/MCTI andthe Mamirauá Sustainable Development Institute – MSDI-OS/MCT. It was completed with support from the School of Natural Re-sources and Environment, the Fisheries and Aquatic Sciences Pro-gram, and the Tropical Conservation and Development Programat the University of Florida; the Society for Marine Mammalogy;and the Associação Amigos do Peixe-Boi. We thank the many peo-ple who contributed to the fieldwork on which this study is based,including the numerous Projeto Boto interns that collected the dataanalyzed. We also extend a special thanks to Dr. Claudia Peñalozaand Dr. William Pine for their valuable modeling advice. ABB wassupported by a National Science Foundation Graduate ResearchFellowship under Grant No. DGE-0802270. The views expressedare those of the authors and do not necessarily reflect the viewsof the above mentioned organizations.

References

Akaike, H., 1973. Information theory and an extension of the maximum likelihoodprinciple. In: Petrov, B.N., Caski, F. (Eds.), Second International Symposium onInformation Theory. Akademiai Kiado, Budapest, pp. 267–281.

Almeida, O.T., Lorenzen, K., McGrath, D.G., 2003. Commercial fishing in the BrazilianAmazon: regional differentiation in fleet characteristics and efficiency. FisheriesManage. Ecol. 10, 109–115.

Barker, R.J., 1997. Joint modeling of live-recapture, tag-resight, and tag-recoverydata. Biometrics 53, 666–677.

Barker, R.J., 1999. Joint analysis of mark-recapture, resighting and ring-recoverydata with age-dependence and marking-effect. Bird Study 46, 82–91.

Barthem, R.B., 1995. Development of commercial fisheries in the Amazon basin andconsequences for fish stocks and subsistence fishing. In: Clüsener-Godt, M.,Sachs, I. (Eds.), Brazilian Perspectives on Sustainable Development of theAmazon Region. UNESCO, Parthenon, Paris, pp. 175–204.

Best, B.C., da Silva, V.M.F., 1989. Biology, status and conservation of Inia geoffrensisin the Amazon and Orinoco river basins. In: Perrin, W.F., Brownell Jr., R.L., Kaiya,Z., Jiankang, L. (Eds.), Biology and Conservation of the River Dolphins –Occasional Paper of the IUCN Species Survival Commission. IUCN, Gland,Switzerland, pp. 23–34.

Best, R.C., da Silva, V.M.F., 1993. Inia geoffrensis. Mammal. Species 426, 1–8.Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference. A

Practical Information–Theoretic Approach, second ed. Springer, New York.Burnham, K.P., Anderson, D.R., 2004. Multimodel inference – understanding AIC and

BIC in model selection. Sociol. Method Res. 33, 261–304.Clapham, P., Van Waerebeek, K., 2007. Bushmeat and bycatch: the sum of the parts.

Mol. Ecol. 16, 2607–2609.Costello, M.J., Baker, C.S., 2011. Who eats sea meat? Expanding human consumption

of marine mammals. Biol. Conserv. 144, 2745–2746.da Silva, V.M.F., 2008. Amazon River dolphin (Inia geoffrensis). In: Perrin, W.F.,

Thewissen, J.G.M., Wursig, B. (Eds.), Encyclopedia of Marine Mammals, seconded. Academic Press, Burlington, MA, USA, pp. 26–28.

da Silva, V.M.F., Best, R.C., 1996. Freshwater dolphin/fisheries interaction in theCentral Amazon (Brazil). Amazoniana 14, 165–175.

da Silva, V.M.F., Martin, A.R., 2000. A study of the boto, or Amazon river dolphin(Inia geoffrensis), in the Mamirauá Reserve, Brazil: operation and techniques. In:Reeves, R.R., Smith, B.D., Kasuya, T. (Eds.), Biology and conservation offreshwater cetaceans in Asia – occasional paper of the IUCN Species SurvivalCommission. IUCN, Gland, Switzerland and Cambridge, UK, pp. 121–131.

da Silva, V.M.F., Martin, A.R., 2010. Status, threats, conservation initiatives andpossible solutions for Inia geoffrensis and Sotalia fluviatilis in Brazil. In: Trujillo,F., Crespo, E., Van Damme, P.A., Usma, J.S. (Eds.), The Action Plan for SouthAmerican River Dolphins 2010–2020. WWF, Fundación Omacha, WDS, WDCS,and SOLAMAC, Bogotá, D.C., Colombia, pp. 123–143.

da Silva, V.M.F., Martin, A.R., do Carmo, N.A.S., 2011. Boto bait – Amazonian fisheriespose threat to elusive dolphin species. IUCN Magazine of the Species SurvivalCommission 53, 10–11.

da Silveira, R., Viana, J.P., 2003. Amazonian crocodilians: a keystone species forecology and management. . . Or simply bait? Crocodile Special. Group Newsl. 22,16–17.

Dillingham, P.W., Fletcher, D., 2008. Estimating the ability of birds to sustainadditional human-caused mortalities using a simple decision rule andallometric relationships. Biol. Conserv. 141, 1783–1792.

Diniz, K.S., 2011. La pesca del bagre zamurito (Calophysus macropterus,SILUFORMES: PIMELODIDADE) y su efecto potencial sobre la extracción detoninas (Inia geoffrensis, CETACEA: INIIDAE) y babas (Caiman crocodilus,CROCODILIA: ALIGATORIDAE) en Venezuela. Masters Thesis, InstitutoVenezolano de Investigaciones Científicas, Caracas, Venezuela.

Estupiñán, G.M.B., Marmontel, M., de Queiroz, H.L., e Souza, P.R., Valsecchi, J., daSilva Batista, G., Barbosa Pereira, S., 2003. A Pesca da Piracatinga (Calophysus

286 V.J. Mintzer et al. / Biological Conservation 158 (2013) 280–286

macropterus) na Reserva de Desenvolvimento Sustentável Mamirauá. RelatórioTécnico. Ministério da Ciência e Tecnologia, Instituto de DesenvolvimentoSustentável Mamirauá, Tefé, Brasil. <http://www.socioambiental.org/website/noticias/agenda/fks/rel_piracatinga.htm> (accessed 31.05.12).

Gómez, C., Trujillo, F., Diazgranados, M.C., Alonso, J., 2008. Capturas dirigidas dedelfines de río en la Amazonia para la pesca de mota (Calophysus macropterus):una problemática regional de gran impacto. In: Trujillo, F., Alonso, J.C.,Diazgranados, M.C., Gómez, C. (Eds.), Fauna acuática amenazada en laAmazonia Colombiana – Análisis y propuestas para su conservación.Fundación Omacha, Fundación Natura, Instituto Sinchi, Corpoamazonia,Bogotá, Colombia, pp. 39–57.

Gómez-Salazár, C., Trujillo, F., Portocarrero-Aya, M., Whitehead, H., 2012.Population, density estimates, and conservation of river dolphins (Inia andSotalia) in the Amazon and Orinoco river basins. Mar. Mammal Sci. 28, 124–153.

Gormley, A.M., Slooten, E., Dawson, S., Barker, R.J., Rayment, W., du Fresne, S.,Brager, S., 2012. First evidence that marine protected areas can work for marinemammals. J. Appl. Ecol. 49, 474–480.

Horton, G.E., Letcher, B.H., 2008. Movement patterns and study area boundaries:influences on survival estimation in capture-mark-recapture studies. Oikos 117,1131–1142.

Jefferson, T.A., Curry, B.E., 1994. A global review of porpoise (Cetacea: Phocoenidae)mortality in gillnets. Biol. Conserv. 67, 167–183.

Leatherwood, S., Reeves, R.R., 1994. River dolphins: a review of activities and plansof the cetacean specialist group. Aquat. Mammal 20, 137–154.

Leatherwood, S., Reeves, R.R., Wuersig, B., Shearn, D., 2000. Habitat preferences ofriver dolphins in the Peruvian Amazon. In: Reeves, R.R., Smith, B.D., Kasuya, T.(Eds.), Biology and freshwater cetaceans in Asia – occasional papers of the IUCNspecies survival commission. IUCN, Gland, Switzerland and Cambridge, UK, pp.131–144.

Loch, C., Marmontel, M., Simões-Lopes, P.C., 2009. Conflicts with fisheries andintentional killing of freshwater dolphins (Cetacea: Odontoceti) in the WesternBrazilian Amazon. Biodivers. Conserv. 18, 3979–3988.

Mangel, J.C., Alfaro-Shigueto, J., Van Waerebeek, K., Cáceres, C., Bearhop, S., Witt,M.J., Godley, B.J., 2010. Small cetacean captures in Peruvian artisanal fisheries:high despite protective legislation. Biol. Conserv. 143, 136–143.

Mansur, R.M., Strindberg, S., Smith, B.D., 2012. Mark-resight abundance and survivalestimation of Indo-Pacific bottlenose dolphins, Tursiops aduncus, in the Swatch-of-No-Ground, Bangladesh. Mar. Mammal Sci. 28, 561–578.

Manzur, M.I., Canto, J., 1997. Las pesquerías de centolla y centollón y suinterferencia con mamíferos marinos. Ambiente Desarrollo 10, 64–69.

Martin, A.R., da Silva, V.M.F., 2004a. Number, seasonal movements, and residencycharacteristics of river dolphins in an Amazonian floodplain lake system. Can. J.Zool. 82, 1307–1315.

Martin, A.R., da Silva, V.M.F., 2004b. River dolphins and flooded forest: seasonalhabitat use and sexual segregation of botos (Inia geoffrensis) in an extremecetacean environment. J. Zool. 263, 295–305.

Martin, A.R., da Silva, V.M.F., 2006. Sexual dimorphism and body scarring in the boto(Amazon River dolphin) Inia geoffrensis. Mar. Mammal Sci. 22, 25–33.

Martin, A.R., da Silva, V.M.F., Salmon, D.L., 2004. Riverine habitat preferences ofbotos (Inia geoffrensis) and tucuxis (Sotalia fluviatilis) in the central Amazon.Mar. Mammal Sci. 20, 189–200.

Martin, A.R., da Silva, V.M.F., Rothery, P.R., 2006. Does radio tagging affect thesurvival or reproduction of small cetaceans? A Test. Mar. Mammal Sci. 22, 17–24.

Matkin, C.O., Durban, J.W., Saulitis, E.L., Russel, D.A., Straley, J.M., Matkin, D.R.,Graeme, M.E., 2012. Contrasting abundance and residency patterns of twosympatric populations of transient killer whales (Orcinus orca) in the northernGulf of Alaska. Fish. Bull. 110, 143–155.

McGuire, T.L., Aliaga-Rossel, E., 2010. Ecology and conservation status of riverdolphins Inia and Sotalia in Peru. In: Trujillo, F., Crespo, E., Van Damme, P.A.,Usma, J.S. (Eds.), The Action Plan for South America River Dolphins 2010–2020.WWF, Fundación Omacha, WDS, WDCS, and SOLAMAC, Bogotá, DC, Colombia,pp. 59–73.

McGuire, T.L., Henningsen, T., 2007. Movement patterns and site fidelity of riverdolphins (Inia geoffrensis and Sotalia fluviatilis) in the Peruvian Amazon asdetermined by photo-identification. Aquat. Mammal 33, 359–367.

McGuire, T.L., Winemiller, K.O., 1998. Occurrence patterns, habitat associations, andpotential prey of the river dolphin, Inia geoffrensis, in the Cinaruco River,Venezuela. Biotropica 30, 625–638.

Northridge, P.S., Hofman, R.J., 1999. Marine mammal interactions with fisheries. In:Twiss, J.R., Reeves, R.R. (Eds.), Conservation and Management of MarineMammals. Smithsonian Institution Press, Washington, DC, pp. 99–119.

Peres, C.A., Lake, I.R., 2003. Extent of nontimber resource extraction in tropicalforests: accessibility to game vertebrates by hunters in the Amazon basin.Conserv. Biol. 17, 521–535.

Peres, C.A., Terborgh, J.W., 1995. Amazonian nature reserves: an analysis of thedefensibility status of existing conservation units and design criteria for thefuture. Conserv. Biol. 9, 34–46.

Perrin, W.F., Donovan, G.P., Barlow, J. (Eds.), 1994. Gillnets and Cetaceans. Rep. Int.Whaling. Comm. (special issue) 15, 629 pp.

Petrere, M., Barthem, R.B., Córdoba, E.A., Gómez, B.C., 2004. Review of the largecatfish fisheries in the upper Amazon and the stock depletion of piraiba

(Brachyplatystoma filamentosum Lichtenstein). Rev. Fish Biol. Fisher. 14, 403–414.

Portocarrero Aya, M., Ferrer, A., Lasso, C.A., Ruiz-Garcia, M., Bolaños-Jiménez, J.,Caballero, S., 2010a. Status, distribution and conservation of the river dolphinsInia geoffrensis and Sotalia spp. in Venezuela. In: Trujillo, F., Crespo, E., VanDamme, P.A., Usma, J.S. (Eds.), The Action Plan for South American RiverDolphins 2010–2020. WWF, Fundación Omacha, WDS, WDCS, and SOLAMAC,Bogotá, DC, Colombia, pp. 17–28.

Portocarrero Aya, M., Hoyt, E., Wood, A., 2010b. Freshwater protected areas andtheir importance for river dolphin conservation. In: Trujillo, F., Crespo, E., VanDamme, P.A., Usma, J.S. (Eds.), The Action Plan for South American RiverDolphins 2010–2020. WWF, Fundación Omacha, WDS, WDCS, and SOLAMAC,Bogotá, DC, Colombia, pp. 159–166.

Read, A.J., 2008. The looming crisis: interactions between marine mammals andfisheries. J. Mammal. 89, 541–548.

Read, A.J., Drinker, P., Northridge, S., 2006. Bycatch of marine mammals in US andglobal fisheries. Conserv. Biol. 20, 163–169.

Reeves, R.R., Leatherwood, S., 1994. Dams and river dolphins – can they coexist.Ambio 23, 172–175.

Reilly, S.B., Barlow, J., 1986. Rates of increase in dolphin population-size. Fish. Bull.84, 527–533.

Robards, M.D., Reeves, R.R., 2011. The global extent and character of marinemammal consumption by humans 1970–2009. Biol. Conserv. 144, 2770–2785.

Rosas, F.C.W., Lehti, K.K., 1996. Nutritional and mercury content of milk of theAmazon river dolphin, Inia geoffrensis. Comp. Biochem. Physiol. 115, 117–119.

Shostell, J.M., Ruiz-Garcia, M., 2010. An introduction to river dolphin species. In:Ruiz-Garcia, M., Shostell, J.M. (Eds.), Biology, Evolution, and Conservation ofRiver Dolphins within South America and Asia. Nova Science Publishers, Inc.,New York, pp. 1–28.

Sinha, R.K., 2002. An alternative to dolphin oil as a fish attractant in the GangesRiver system: conservation of the Ganges River dolphin. Biol. Conserv. 107,253–257.

Smith, A.M., Smith, B.D., 1998. Review of status and threats to river cetaceans andrecommendations for their conservation. Environ. Rev. 6, 189–206.

Speakman, T.R., Lane, S.M., Schwacke, L.H., Fair, P.A., Zolman, E.S., 2010. Mark-recapture estimates of seasonal abundance and survivorship for bottlenosedolphins (Tursiops truncatus) near Charleston, South Carolina, USA. J. Cetac. Res.Manage. 11, 153–162.

Taper, M.L., Staples, D.F., Shepard, B.B., 2008. Model structure adequacy analysis:selecting models on the basis of their ability to answer scientific questions.Synthese 163, 357–370.

Trujillo, F., Gómez, C., Portocarrero, M., Diazgranados, M.C., 2007. Especies de Fauna:mamíferos acuáticos. In: Ruiz, S.L., Sánchez, E., Tabares, E. Prieto, A., Arias, J.C.,Gómez, R., Castellanos, D., García, P., Rodríguez, L. (Eds.), Diversidad biológica ycultural del sur de la Amazonia colombiana. Corpoamazonia, InstitutoHumboldt, Instituto Sinchi, UAESPNN, Bogotá, Colombia, pp. 439–440.

Trujillo, F., Crespo, E., Van Damme, P., da Silva, V., Usma, S., Aliaga, E., Gómez-Salazár, C., Portocarrero, M., Suárez, E., Ferrer, A., Bolaños, J., 2010a. The actionplan for river dolphins in South America. In: Trujillo, F., Crespo, E., Van Damme,P.A., Usma, J.S. (Eds.), The Action Plan for South American River Dolphins 2010–2020. WWF, Fundación Omacha, WDS, WDCS, and SOLAMAC, Bogotá, DC,Colombia, pp. 179–198.

Trujillo, F., Crespo, E., Van Damme, P., Usma, S., Morales-Betancourt, D., Wood, A.,Portocarrero, M., 2010b. Summary of threats for river dolphins in SouthAmerica: past, present and future. In: Trujillo, F., Crespo, E., Van Damme, P.A.,Usma, J.S. (Eds.), The Action Plan for South American River Dolphins 2010–2020.WWF, Fundación Omacha, WDS, WDCS, and SOLAMAC, Bogotá, DC, Colombia,pp. 147–158.

Trujillo, F., Portocarrero-Aya, M., Gómez-Salazár, C., Diazgranados, M.C.,Castellanos-Mora, L., Ruiz-García, M., Caballero, S., 2010c. Conservation statusof river dolphins Inia geoffrensis and Sotalia fluviatilis in the Amazon and OrinocoRiver Basins in Colombia, in:Trujillo, F., Crespo, E., Van Damme, P.A., Usma, J.S.(Eds.), The action plan for South America river dolphins 2010–2020. WWF,Fundación Omacha, WDS, WDCS, and SOLAMAC, Bogotá, D.C., Colombia, pp. 29–57.

Turvey, S., Pitman, R.L., Taylor, B.L., Barlow, J., Akamatsu, T., Barrett, L.A., Zhao, X.,Reeves, R.R., Stewart, B.S., Wang, K., Wei, Z., Zhang, X., Pusser, L.T., Richlen, M.,Brandon, J.R., Wang, D., 2007. First human-caused extinction of a cetaceanspecies? Biol. Lett. 3, 537–540.

Utreras, V., Suárez, E., Salime, J., 2010. Inia geoffrensis and Sotalia fluviatilis: A briefreview of the ecology and conservation status of river dolphins in theEcuadorian Amazon. In: Trujillo, F., Crespo, E., Van Damme, P.A., Usma, J.S.(Eds.), The Action Plan for South American River Dolphins 2010–2020. WWF,Fundación Omacha, WDS, WDCS, and SOLAMAC, Bogotá, DC, Colombia, pp. 87–97.

Vidal, O., 1993. Aquatic mammal conservation in Latin America: problems andperspectives. Conserv. Biol. 7, 788–795.

Wade, P.R., 1998. Calculating limits to the allowable human-caused mortality ofcetaceans and pinnipeds. Mar. Mammal Sci. 14, 1–37.

White, G.C., Burnham, K.P., 1999. Program MARK: survival estimation frompopulations of marked animals. Bird Study 46, 120–139.