Content uploaded by Oliver C. Stringham
Author content
All content in this area was uploaded by Oliver C. Stringham on Sep 22, 2018
Content may be subject to copyright.
J Appl Ecol. 20 18 ;1–9. wileyonlinelibrary.com/journal/jpe
|
1
© 2018 The Aut hors. J ournal of Applie d Ecolog y
© 2018 British Ecological Society
Received:30April2018
|
Accepted:24June2018
DOI :10.1111/1365-2664.13237
RESEARCH ARTICLE
Pet problems: Biological and economic factors that influence
the release of alien reptiles and amphibians by pet owners
Oliver C. Stringham | Julie L. Lockwood
Depar tmentofEcology,Evolution,and
NaturalResources,RutgersUniversit y,New
Brunswick,NewJersey
Correspondence
OliverC .Stringham,DepartmentofEcology,
Evolution,andNaturalResources,Rutgers
University,14CollegeFarmRoad,N ew
Brunswick,NJ0 8901.
Email:oliverstringham@gmail.com
HandlingEditor:CélineBellard
Abstract
1. Thenumberofalienreptilesandamphibiansintroducedandestablished world-
widehasincreasedoverthelastdecades.Thelegalpettradeisnowthedominant
pathway by which individuals of these species arrive in their non-nativelocale.
Despite itsimportance, specificfactorsofpettradepathwaythatinfluencethe
release (introduction) of exotic reptiles and amphibians have not yet been
examined.
2. Wesetout to identify broadscale andeasily measuredbiologicaland economic
factorsthat influence the release of these exoticpets bytheir owners. We hy-
pothesize that b iological fac tors reflect t he cost of care, and e conomic facto rs
reflectthevaluethatownersplaceontheirpet,bothofwhichcaninfluencethe
probabilitywhenapetisreleased.Wecollectedlifehistoryandeconomicdataon
the1,722speciesofreptilesandamphibianssoldwithintheUSaspetsoverthe
last 18years. We alsocompiled a listofpettrade-attributedreleasesinthe US
(i.e., all free-living species regardless of whetherthey successfullyestablished).
Weusedboosted regression treestocorrelate species releasestatus withtheir
life-historytraitsandeconomicattributes(r2=0.51,AUC=0.89).
3. Wefoundthatspecieswithahighprobabilityofbeingreleasedwereimportedat
higherquantitiesoverourperiodofrecord,havearelativelylargeadultmassand
commandedcheaperretailprices.Thenumberimportedandpriceinteractedwith
longevit y and adult mass to produ ce nonlinear incre ases in release prob ability.
The most important interaction revealed that large-bodied species imported in
highquantitieshave athree timeshigherreleaseprobabilitycompared tolarge-
bodiedspeciesimportedinlowerquantities.
4. Policy implications.Ourresultsprovideguidancetowardstargetingexoticpetrep-
tileandamphibianspeciesthatareatahighriskofbeingreleased.Speciesthatare
bothprevalentinthe pettrade andlarge-bodiedor long-livedhavethe highest
probab ilityofbe ingreleas ed.Thiswillaidind evelopinged ucationandpolicysolu-
tionsaimedatdecreasingtherateatwhichthesepetsarereleased,thuscurtailing
theinvasionprocessbeforethesespeciescanestablishandimpactscanoccur.
KEYWORDS
alienspecies,amphibians,exoticpettrade,introductionprobability,introductionstage,
invasions,petreleases,reptiles
2
|
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
1 | INTRODUCTION
Invasive reptiles and amphibians impose severe ecological and
socio-economic costs including sometimes driving entire suites of
nativespeciestowardsextinction(Shine,2010;Wiles,Bart,Beck,&
Aguon,2003).Overthepasttwodecades,thetradein“exotic”pets
hasbecomethemainsource ofalienreptileandamphibian species
worldwide(Capinha etal., 2017;Kraus, 2009; Kr ysko etal.,2011).
Exoticpetsarespecieswithoutalonghistoryofdomesticationthat
arelegallycapturedfromtheirnativerangeorbredwithin facilities
andsoldtoconsumersashouseholdcompanions.Individuals sold
asexoticpets are neverintended for release; however, the regular
observationoftradedspeciesasfree-livingsuggeststhatanontriv-
ial fraction is released (Hulme, 2015). These released individuals
thenhavetheopportunitytoestablishalocalpopulation,withsome
further probability ofanestablished populationimposingnegative
impact s on co-occurring native species (Blackburn etal., 2011).
Despitereleasebeingtheinitiatingstepofinvasions,weknowvery
little about theprocess by which exotic pets are released,in large
partbecausepetreleaseeventsaregeographicallylocalisedandare
knowntobe rare relativeto thenumberofpeople who ownexotic
pets (Strecker,Campbell, & Olden,2011).Addto this thedifficulty
ofdocumentingthe dynamiccomposition ofthe exotic petmarket
(Romagosa,2014),anditbecomes clear thatthetaskof connecting
trade,specieslifehistoryandconsumerbehaviourtotheoccurrence
offree-livingpetreptilesandamphibiansisaformidablechallenge.
Here,wegeneratehypothesesabouthowthesespecies’biological
traits, and the economic factors related to their trade, influence
the release decisions of pet owners. We testthese hypothesesby
combining pet market data with in-depth records of documented
releasedfree-livingreptilesandamphibianswithintheUnitedStates
(US).
Researchoninvasionvec torsandpathwaysmakesclearthatthe
suite of species transported via human actionsis set bythe inter-
playofsocio-economicfactorsandspecies’biology(Essletal.,2011;
Hulmeetal.,2008).Oneoftheprimaryfilters(orbarriers)thatalien
speciesmusttransitearlyintheinvasionprocessisbeingreleasedas
free-livingwithin a non-native environment.Inthepettradepath-
way,release occurs eitherwhen the wholesaler or consumer fails
to keep the in dividual in a se cure enclosu re (escape) or wh en the
owner (a consumer or the wholesaler) purposefully removes the
individual from captivit y and allows it to become free-living (Su,
Cassey, & Blac kburn, 2016; Vall-llos era, Woolnough , Anderson , &
Cassey, 2017). Since the pettradepathway is quite distinct in the
degreetowhichhumanbehaviourinfluencesthereleaseeventitself,
weshouldexpectthatdriversofreleasearespecifictothispathway
(Hulme etal., 2008).Forexample,smaller-bodiedcargohitchhikers
may be more li kely to be release d (accidental ly) because t hey are
morelikelytogoundetected.However,thereverserelationshipmay
betrueforthepettradepathwaygiventhatlarger-bodiedspecies
tendtobereleasedmoreoftenduetothemoutgrowingtheirhous-
ing(termed “tankbusters”forfish:Holmbergetal.,2015).Thus, in-
dependentanalysesofreleasepathwaysareimperativetoproducing
meaningfulinsightsforpolicydevelopment(Hulmeetal.,2017).
Despite the impor tance releases play in the invasion process
forthepettradepathway,mostoftheresearchtodatehasfocused
on the factors influencing the establishment ofexotic petpopula-
tions and not onthefactors relatedtotheir initialintroduction (or
release; B omford, Kraus , Barry, & Lawrence , 2009; Fujisaki eta l.,
2010;Mahoney etal.,2015; VanWilgen & Richardson,2012).The
Name Description Sample size
Expected effect on
release probability
Lifehistory
Adultmass Meanbodymassofadults(g) 1 ,115 ↑
Longevity Maximumnumberofyears
knowntosurviveeitherin
wildorincaptivity
909 ↑
Reproductive
output
Numberofeggsperyear,
calculatedbymultiplying
clutchsizebyclutch
frequency
748 ↑
Economic
Quantit y
imported
Totalnumberoflive
individualsimportedintothe
US(designatedforthepet
trade)from1999to2013
1,426 ↑
Price Five-yearmedianoftheretail
priceofaspecies(US$)
627 ↓
Timeonmarket Numberofyearsthata
specieswasavailablefor
saleonthepetmarket(from
1to18years)
1,722 ↑
TABLE1 Biologicalandeconomic
factorshypothesizedtoinfluencethe
probabilitythatapetreptileoramphibian
willbereleasedasfree-livingbytheir
owners.Samplesizerepresentsthe
numberofspeciesthathavedataforthe
variable(outofatotalof1,722species)
|
3
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
few published studies on pet species release are lacking in both
taxonomicandanalyticalbreadth (e.g., Duggan,Rixon,&MacIsaac,
2006; García-Díaz & Cassey, 2014; Su etal., 2016; Tingley etal.,
2010).Consequently,the releasestage isfrequentlyignored inrisk
assessment(Leung etal., 2012), eventhough there is considerable
evidence t hat if alien speci es are not release d, or are release d in
small enough numbers, they are far less likely to become estab-
lished (Cassey, Delean, Lockwood, Sadowski, & Blackburn, 2018;
Lockwood,Cassey,&Blackburn,2005).
Hereweseektoclosethisresearch gapbyexploring theability
of biologic al and econom ic drivers to exp lain releases of p et rep-
tilesandamphibians(Table1).Forbiologicaldrivers,wehypothesize
thattheprobabilityofreleasewillincreasewhenthereisamismatch
betweentheperceivedandactual level ofcareneeded tomaintain
theindividualincaptivity.Weexpect thatwell-meaningconsumers
purchase apetbutfinditdifficultto carefor itoverthelongterm,
andthatsomefraction oftheseownerswillchoosetoreleasetheir
petratherthanselloreuthanizeit.Weposit that care costswillbe
higherforspeciesthatgrowtolargeadultsizes,liveforalong time
andarecapableofproducingmanyyoungwhileincaptivity.Foreco-
nomicdrivers,wehypothesizethatthe more abundantandlessex-
pensiveaspecieswithinthepetmarket,andthelonger thespecies
remainsonthemarket,themorelikelytheindividualsofthisspecies
aretobereleased.Ownersmayplacelessvalueonlow-costpetsand
thereforebemorelikelytoreleasethemwhencarebecomescostly
or inconveni ent. Furth ermore, all e lse being equa l, the proba bility
thatatleastoneindividualofan exotic petspecieswillbereleased
will incre ase if the specie s is commonly sol d to consumers, s ome
fract ion of which will r esort to rel ease when the p et is no longer
wanted.Byattemptingtoquantifythefactorsrelatedtotherelease
ofexotic pets, we seek to help guide effortsaimed at curbing pet
animalreleases.
2 | MATERIALS AND METHODS
2.1 | Pool of species available for purchase as pets
We identifi ed the range of sp ecies that ar e available for p urchase
aspetsintheUSbycombiningtwodatasources.Thefirstisimport
recordskeptbytheUSFishandWildlifeSer vice(LawEnforcement
ManagementInformation System, LEMIS;seeRomagosa, 2014for
moredetails).Thesedataarecollectedbytrainedinspectionagents
atairportsacrosstheUS.Wefocusedouranalysisonthecontermi-
nousUSandthereforeexcludedimportrecordsfromHawaii,Alaska
and US terr itories. LEMIS da ta from 1999 to 2016 indicate d that
1,613reptileandamphibian specieswereimportedintotheUS for
thepettrade.
Our second data source was alist of species available for pur-
chaseonhighlytraffickednationwideinternetvendorsfrom2012to
2016.Weexpectedthisinventorytooverlapsubstantiallyinspecies
composition with the LEMISdata.However, the web retailer sites
also include species thatare breddomesticallyand thusserves to
expandourpoolofspeciesthathadtheopportunitytobereleased
bypetowners.Thevendorswechosetoincludeinoursurveyarethe
topthreemosttraffickedreptile andamphibianinternet-basedpet
stores,eachofwhichofferstoshipindividualpetstoanyconsumer
intheconterminousUS.Fromtheseonlinepetretailers,wecollated
94,230 uniqueindividualpetlistingsrepresenting 652species (see
Suppor ting Information DataS1 for details onour web-based data
collectionprotocol).
2.2 | Variables that influence release
Next, we gathered information on biological factors that we hy-
pothesizedto berelatedto releaseprobabilit y(explanatory),which
include d mean adult bo dy size (gram s),m ean reprodu ctive outp ut
(eggs/year)andmaximumlongevity(years;Table1).Weexpectthat
as each of the se biologic al factor s increase s in magnitud e, the in-
curredcostsofcarewillalsoincrease,thusincreasingreleaseprob-
ability.Wecollectedthisinformationfromfivepublisheddatabases
ofvertebratelife-historytrait s(seeSupportingInformationDataS2
fordetailsoncompilingdatabases).Additionally,followingHolmberg
etal. (2015) we com pared the diffe rence between age a nd body
mass at time of s ale (retail size and age) wit h maximum age and
adultbodymassforspeciesappearingonthewebvendorsites(see
SupplementalInformationDataS4fordetailsonestimatingageand
massatsale).Datacoverageformaximumlongevityandadultmass
wassubs tantiallyhigherthanretailmassandage,andthetwomeas-
ureswerehighlycorrelated(Suppor tingInformationDataS3).Thus,
weonlyincludedmaximumlongevity andadult massperspeciesin
theboostedregressiontree(BRT)modeldescribedbelow.
Fortheeconomicexplanatoryfactors,wecalculatedthe length
oftimethateachspeciesappearedintheUS petmarketbytallying
thenumberofyearseachspecieswaslistedatleastonceacrossany
ofthethreewebsitesand/orwithintheLEMISdata. Wecalculated
thenumberofimportsforeachspeciesbysummingacrossallyears
the numb er of live individ uals designate d for the pet tr ade in the
LEMISdata (availablefor1999–2013).We also estimated the price
that an individualof each species commands on the retail market.
Wecollatedpriceinformationfromallweblistingsperspecies,cor-
rected a ll prices to rep resent Dece mber 2016 US$ and ca lculated
themedianofthisdistribution(SupportingInformationDataS1).
2.3 | Testing for release patterns
We compiled a lis t of documented r eleased (also k nown as intro-
ducedor free-living)exoticreptiles andamphibiansprimarily using
Kraus (2009),whichis the mostcomprehensivedatasetofreleases
available onthese taxa for the US.Weonly included species from
Kraus (2 009) if he liste d them as having a rrived in the U S via the
pettrade. Correspondingto thetemporalspanof ourspeciespool,
weonlyrecordedas“released”speciesthatKrausindicates ashav-
ingbeen observedasfree-living after1999.BecauseKraus(2009)
only had a ccess to information u p to about 2008 , and to ensure
comprehensiveaccountingofreleasedspecies,weincludedspecies
4
|
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
recordedasfree-livingfrom1999to2016withinEDDMapS (2012;
www.eddmaps.org).EDDMapSisacitizenscienceefforttoprovide
real-time in formation on ali en species occ urrences acro ss the US,
whereeachlistingisverifiedbytaxonomicexperts.Therewere114
morespeciesrecordedasreleasedfromEDDMapSthatwerenotin
Kraus(2009)withinourperiodofrecord.Weexcludedallrecorded
releasesfrom Hawaii and Alaska tomatch ourgeographical scope
ofspeciesavailability(above).Notethatweincludedspeciesas“re-
leased”regardlessofwhethertheyhavenow,oreverwill,establish
self-sustainingpopulationsintheUS.Ourinterestiswiththesetof
petspeciesthatwerereleasedorescapedfromcaptivityregardless
oftheir long-termfate once free-living.This effort resulted in 126
speciesrecordedasfree-livingsomewhereintheconterminousUS
betwee n 1999 and 2016. We designate d these free -li ving specie s
as“released”in our analyses below,andthe species that werenot
recordedas free-livingto be “not-released”(our binary dependent
variable).
Weused BRT toexplore the relationshipbetweenourexplana-
toryvariablesandthereleasestatusofreptilesandamphibianssold
aspetsintheconterminousUS.BRTisamachinelearningtechnique
that can fit complex nonlinear relationships and handle different
typesofdata allwhile potentiallyyieldinghigherpredictiveperfor-
mancethantraditionalstatisticalmodels(Elith,Leathwick,&Hastie,
2008;seeSuppor tingInformationDataS5forBRTparametersused
and testing of simpler models). Our reasonfor choosing BRT over
more tra ditional st atistical me thods such as GL Ms is its abilit y to
accommodateformissingdata. Whileweusedthemostup-to-date
databases (to our know ledge) to compile life-history t raits, there
were still missing data for all traits (Table1). The BRT algorithm
handles missingdata byskippingthenodewithmissingdataduring
treebuilding(i.e.,dataarenotimputed,butrathernotincluded;Elith
etal., 20 08). BRTs were mode lled in r v. 3.3.1 using t he packages
gbm v. 2.1. and dismo v. 1.1.
Toassessphylogeneticcorrelationsbetween species(i.e.,Allen,
Street, & Capellini, 2017;Capellini,Baker,Allen, Street,&Venditti,
2015),weperformed twoadditionalBRTmodelsthatincluded tax-
onomy(FamilyandOrder)asaproxyfor phylogeny(i.e.,Schmidt&
Drake,2011;VanWilgen&Richardson,2012).Thisanalysisdoesnot
directlymeasurephylogeneticcorrelations(e.g.,Pagel’sλ),whichisa
shortcomingofmachinelearningtechniquesingeneral(Supporting
InformationDataS6).Thus,werecommendthatmethodsbedevel-
opedtoincorporatephylogeneticinformationintomachinelearning
techniques including BRTs.With this caveat, from our analysis we
foundlittle evidenceforaphylogeneticsignal inreleaseprobability
(Suppor tingInformationDataS6),andgiventhattaxonomyorphy-
logenyalonedoesnotrevealwhichlife-historytraitsareimpor tant
predictorsofrelease,wefocusourinterpretationofresultsbelowon
theBRTwithouttaxonomy.
ToevaluatetheperformanceandfittheBRTmodel,weusedtwo
metrics.First, we calculated the areaunder thereceiver operating
characteristiccur ve(AUC)for boththetraining(alldata)andcross-
validatedmodels(averageofallcross-validateddatafolds).Second,
wecalculated r2 (percentdeviance explained)forboth the training
andcross-validatedmodels.
To explore how each explanatory variable influenced release
probability,weexaminedfourdifferentmetrics.First,wecalculated
therelativecontributionmetricforeachexplanatoryvariable,calcu-
latedasthenumberoftimesavariableisselectedforsplittingatree,
weighted by the improvement result ing from that split , averaged
over all tree s. Second, we gene rated partia l dependency p lots to
visualizetherelationshipbetweeneachexplanatoryvariableandre-
leaseprobability.Third,foreachexplanatoryvariable,wemeasured
the magnitude of its influence on release probability,calculated as
therange (max–min) ofthe valuesinthepartialdependency plots.
Wetermed this value ∆P(release), which is essentiallya measure
ofeffectsize.Finally,weexamined therelative strengthofvariable
interactionsbycalculatingtheinteractionsizebetweenexplanatory
variables(valuesnearzerorepresentnegligibleinteractions)andvi-
sualizedtheseinteractionsinheatmaps(Elithetal.,2008).
3 | RESULTS
Ofthe1,722 speciesrecordedeither in LEMIS or on the websites,
Lacertilia(lizards)werethegroupwiththemostspeciesforsale(739
species),followedbySerpentes(snakes)with490species(Figure1a).
Over half of t he species (1,106) were availa ble on the market f or
≤5years (Fig ure1b), while ~10% of species were av ailable for th e
FIGURE1 (a)Speciesforsaleinthe
USexoticpetmarketbytaxonomicgroup.
Thegroupsinorderfromlefttorightare:
Anura,Caudata,Crocodylia,Testudines,
LacertiliaandSerpentes.Thepercentages
ofspeciesineachtaxonomicgroup
recordedasreleasedarelistedabove
eachbar.(b)Numberofyearsspecies
wereavailableforpurchasewithintheUS
exoticpetmarket
0
200
400
600
800
Number of species
0
100
200
300
400
024681012141618
Years on market
Number of specie
s
(a)
(b)
4.6%
6.7% 36.4%
7.1%
9.5%
4.9%
|
5
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
fulldurationofthisanalysis,18years.Mostweblistingssoldpetsfor
under$100althoughsomespeciescommandedsubstantiallyhigher
prices(SupportingInformationFigureS1).Foralltaxa,theageatthe
timeofsale(retailage)wasmarkedlylessthanthespecieslongevity
(Figure2a). A similar trendwas obser ved for retail mass andmaxi-
mumadultmass(Figure2b).
TheBRTwaseffectiveatexplainingreleasestatus,withacross-
validatedAUCscoreof0.89andar2of0.51.Asanticipated,theAUC
and r2aresomewhathigherforthetrainingdata(AUCof0.94andr2
of0.66),sinceoverfittingislikelywithoutcross-validation.
Three variables had the largest effect on release probability:
quantityimported,priceandadultmasswitha∆P(release)of48,23
and21percentage pointsrespectively.Intermsofmodel fit,these
samevariableshadthelargestrelativecontributiontoreleaseproba-
bility(ranging17%–30%;Table2).Thethreeotherfactorsexplored,
time on market, longevity and reproductiveoutput, had less of an
effectonreleaseprobabilityandmodelfit.
Partial dependency plots showed that quantity imported, lon-
gevity, reproductive output and time on market had a positive
relationship with release probability,whereasprice had anegative
relationship(Figure3). Adultmasshadan unequal“U”-shapedrela-
tionshipwithreleaseprobability,whereintermediatevaluessawthe
FIGURE2 Comparisonofretailage/
size(verticallines)andlongevity/adult
size(histograms)foreachtaxa.(a;left
column)Longevityofspeciessoldonthe
USexoticpetmarketbytaxonomicgroup.
(b;rightcolumn)Adultmassofspecies
soldontheexoticpettradebytaxonomic
group.Verticalsolidlinesindicatethe
medianretailage(a)orretailmass(b)of
allspeciesinthespecifiedgroup.Dashed
verticallinesrepresentthe5thand95th
percentilesofretailage(a)orretailmass
(b).NoteforCaudata,therewerenodata
forretailage,andforCrocodyliatheretail
masswasrecordedforonlyonespecies,
hencenovisiblepercentilelines
0
40
0
10
0
20
0
4
8
0
100
200
0
50
100
10 30 50 70+
0
10
20
0
10
0
30
60
0
4
8
0
100
0
50
100
–1 012345
0
40
0
10
0
20
0
4
8
0
100
200
0
50
100
10 30 50 70+
0
10
20
0
10
0
30
60
0
4
8
0
100
0
50
100
10 12345
Longevity (years)
log10adult mass (g)
Number of species
(a) (b)
TABLE2 Resultsfromboostedregressiontreemodel.△ P
(release)isameasureoftheeffectsizeofthevariableandis
calculatedastherangeofthepar tialdependencyplots(Figure4).
Relativecontributionisameasureofvariablefitandisscaledso
thatthecontributionsofallvariablesaddsupto100% ,where
highernumbersindicatestrongercontributions
Variable △ P (release)
Relative
contribution (%)
Quantityimported 0.48 29. 7
Price 0.23 22 .1
Adultmass 0.21 17. 2
Timeonmarket 0.10 8.5
Longevity 0.09 11.5
Reproductive 0.04 11.0
6
|
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
lowest rel ease proba bility, while la rger massed sp ecies had hi gher
release probability compared to smaller massed ones(Figure3). In
termsofvariableinteractions,thestrongestwasbetweenadultmass
and quantity imported(interaction sizeof56),where large-bodied
species imported in high quantities have three times higher re-
lease probability compared to large-bodied species imported in
lower quantities (Figure4a). There was a less strong but similar
relationship between quantity imported and longevity whereby
FIGURE4 Interactionsinourboostedregressiontree(BRT)modelexplainingreleaseprobabilityofexoticpetreptilesandamphibians
intheUS,including(a)quantityimportedandadultmass,(b)priceandadultmassand(c)quantityimportedandlongevity.Contour
lineswithassociatednumbersrepresentfactorof10%increasesinreleaseprobability.Darkershadesindicatehigherprobabilityof
releaseaspredictedfromtheBRTmodel.Interactionsizesforeachinteractionare56(a),16(b)and4(c)
log 10 adult mass (g)
log10quantityimported
2
4
6
(a) (b) (c)
1234
0.1
0.2
0.4
0.6
0.7
0.8
0.3
0.5
1.0
1.5
2.0
2.5
3.0
1234
0.1
0.2
0.2 0.4
0.4
0.5
2
4
6
10 20 30 40 50 60
0.1
0.2 0.3
0.4
0.0
0.2
0.4
0.6
0.8
1.0
Longevity (years)
log 10 adult mass (g)
0.7
0.6
log10median price(US$)
log10quantityimported
P(release)
FIGURE3 Partialdependencyplotsshowingeffectsofbiologicalandeconomicfactorsonreleaseprobabilityforexoticpetreptilesand
amphibiansintheUSgeneratedfromourboostedregressiontreeanalysis.Partialdependencyplotsshowtheeffectofagivenexplanatory
variableonreleaseprobability,whileholdingtheeffectsofotherexplanatoryvariablesattheiraverage.Weused500bootstrapreplicatesto
calculatethe95%confidenceintervals,depictedingrey.Notethatseveralofthex-axesareonalog10scale.Acrossthetopofeachplotisa
rugplotshowingthedistributionofdatapoints
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
1100 10000 1000000
Quantity imported
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
10 1001000
Price (US$)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.01 1100
Adult mass (kg)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
050 100 150
Longevity (years)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
10 1000
100
Reproduction (eggs/year)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
51015
Time on market (years)
Release probability
|
7
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
long-lived species im ported in large quantity h ave higher release
probabilitycomparedtolong-livedspecies importedatlowquanti-
ties(Figure4c).Finally, weshowthat adultmassandpriceinteract
sothatlowpriceandlargerelativeadultmasscombinetoproducea
substantialriseinprobabilityofrelease(Figure4b).
4 | DISCUSSION
The pet tr ade has taken o n increased im portan ce in conserv ation
with regard to invasive species (Smith etal., 2009), as is evident
from the w idespread eco logical and e conomic damage t hat once-
pet specieslike Burmese python (Python bivittatus) and Pacific li-
onfish (Pterois volitans)can produce(Dorcas etal.,2012;Whitfield,
Gardner,Vives,&Hare,2002).Here,wesetouttoexaminetherole
ofbiologicalandeconomicfactorsinexplainingthefirststepinthe
reptileandamphibianinvasionpathway,releaseprobability.Ourre-
sults indicate that there areclear biological and economicfactors
thatincreasetheprobabilitythatexoticpetreptilesandamphibians
will be obs erved as fre e-li ving (but not ne cessarily es tablishe d) in
theconterminousUS.Thesefactorsreflectthecostofpetcarethat
ownersbearoverthespanofthepet’slifeandthevaluethatthese
ownersplaceonthepet’sthemselves.Notably,wedemonstratethat
consideringbothlife-historytraits and economic characteristics is
necessarytodeterminereleaseprobabilityasthetwofactorsoften
interacttoelevatereleaseprobabilit y.Thisknowledgeallowsforthe
proactivedevelopmentofpolicyandeducationtoolsthatcanserve
todecreaseeithertheprevalenceoflikelyreleasedexoticpetsinthe
marketandinhomesorencourageownerstoresponsiblydisposeof
unwantedpets.
The set of sp ecies we record a s released ar e the produc t of a
complexset of individualownerdecisions. Our approachseeksto
identify broadscale and easily measured factors that effectively
captureand describe these decisions. In theabsenceofextensive
pet owner surveys,our approach serves asareasonable first step
toelucidatingmechanismsthatdriveexoticpet releases.Wefound
thatreptilesandamphibiansthataremorea bundantonthepetmar-
ket and are so ld at cheaper pr ices are more like ly to be release d.
Thisresultisconsistentwithpreviousresearchonreleasesthatcon-
cludes thatpetabundanceisa majorfactor for explainingreleases
inother exoticpet groups(Dugganetal., 20 06; García-Díaz,Ross,
Ayres,&Cassey,2015). Our findingthatlargeadultmassincreases
release probability is also consistent with previous research for
othertaxonomicgroups(Holmbergetal.,2015;Tingleyetal.,2010).
Therefore,itislikelythatlargerspeciesmaybeaparticularcausefor
concern forreleasesamong most exotic vertebratepet taxa(birds
mightbeanexception:Su, Cassey,Vall-llosera,&Blackburn,2015).
Furthermore, we showthatadult massandlongevit yinteract with
economic factors to either diminish or augment release probabil-
ity. For inst ance, long-lived s pecies that are p ervasive on t he pet
market are ove r three times m ore likely to be rele ased compare d
to less commonly soldspe ciesof comparable longevity. Similarly,
species thatachieveahighadult mass whilealsobeing commonin
themarketandlow-pricedaremuchmorelikelytobereleasedthan
speciesofsimilarmassthatarerareonthemarketorexpensive.One
caveatofmachinelearningmethodsisthattheydonotallowexplicit
testingforphylogeneticsignal,whichisashortcomingtoourresults.
Phylogeny wasfound to be importantinaseparatestudy onalien
reptilesandamphibians(species not uniquetothepettrade; Allen
etal.,2017),suggestingthatthesamemaybetruewithrelease.
Cross-stage differences in the importance of life history and
socio-economic factors are a common feature across all biological
invasions(Lockwood,2017)andourresultsfurtherhighlighttheim-
portanceofcompartmentalisingeachstageof the invasionprocess
when quan tifying risk (B lackburn etal., 2011; Leun g etal., 2012).
For insta nce, we show that lar ger-bodie d reptiles and a mphibians
aremorelikelytobereleased;however,afterrelease,otherresearch
suggests that smaller-bodied species are more likely to establish
self-sustainingpopulations(fromglobalestablishments:Allenetal.,
2017;Mahoneyetal., 2015). Similarly,our results suggest thatim-
portquantitystronglyincreasesreleaseprobabilityforreptilesand
amphibiansbutFujisakietal.(2010;Floridaonly)indicatethissame
variabledoesnottranslateinto higherprobability ofsuccessfules-
tablishmentinthesesametaxa.Similarly,thereisnoreasontosus-
pect thattrait sthatareknownto elevateestablishmentsuccess or
invasivespread,suchashabitatbreadthorcompetitiveability(Van
Wilgen&Richardson,2012),willhaveastronginfluenceonrelease
probability.
Current management options aimedatreducingexotic pet re-
leases in clude educati ng the pet-own ing public of the eco logical
risks of releasing their pets (Reaser & Meyers, 2007), or imple-
mentation by local agencies of “buy backs” and“amnest y events”
forownersseeking to relinquishtheir pet (Hardin, 20 07). Ourre-
sults provideaninitial evidence base to allowsuch approaches to
target p articular sp ecies, and the ir owners, mak ing these effo rts
more eff ective and ef ficient in their m ission. Specif ically, target-
ingspecies thatarecommonly soldandlarge-bodied orlong-lived
will have the g reatest eff ect in reduci ng the number of r eleases.
Warning consumers about the zoonotic risks of exotic pets has
been shown to decrease the probability when apet is purchased
(Moorhouse,Balaskas,D’Cruze,&Macdonald,2017).Wetherefore
sugges t that a similar st rategy of maki ng resources a nd informa-
tionavailableabouthowa pet’sadultsizeandlongevityincreases
carerequirementscouldhelpsteerconsumersawayfrompurchas-
ing reptil es and amphibi ans they are likely to r elease after i nitial
purchase (Seekamp, Mayer,Charlebois,&Hitzroth,2016).Inaddi-
tion,tofurtherhelppreventconsumersfromreleasing theirexotic
pets, i t could be bene ficial if ret ailers infor med consume rs about
safeplacestosurrendertheirpetssuchexoticpetshelters,amnesty
eventsorreturns/buybackstotheretailers.
Integrat ing the release st age into risk managem ent can result
in a more robust and accurate assessment of invasion risk (e.g.,
Leung & Man drak, 20 07). Our approa ch utilizes infor mation com-
montoexistinginvasionriskassessments(e.g.,life-historytraits),or
canbeeasilyaquired (e.g., currentpriceorpopularit y),allowing for
easier an d immediate integ ration of relea se risk facto rs. Such risk
8
|
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
assessm ents have bee n used to guide l egislation ai med at curbin g
invasionsthroughimportbansofhighriskspecies(e.g.,throughthe
Lacey ActintheUS:Fowler,Lodge, & Hsia, 20 07). Our results are
particularlyhelpfulinthiscontextasthediversityofreptileandam-
phibian sp ecies sold a s pets has be en steadil y increasing ove r the
lastt wodecades (Romagosa,2014).Finally,ourresultscanbe used
tocraftlegislationtargetedatreducingtheprobabilityofreleaseof
species.Forexample,ourresultscanbeusedtotargettaxingandli-
censingeffortstowardshigh-releaseriskspecies,requiringretailers
toonlysellsingle-genderedindividualsofahigh-releaseriskspecies,
and“tagging”(viamicrochips)individualsfromhigh-releaseriskspe-
cies. Regardlessoftheapproach,adata-drivenef fortto document
factorsthatresultinexoticpetreleasescanadvanceamorecompre-
hensive,evidence-based approach to riskmanagement and policy
implementation.
ACKNOWLEDGEMENTS
Wethank Nick Mandrak, PhillCassey,MalinPinsky, BrookeMaslo
andtheLockwoodLab(RutgersUniversity)forhelpful insightsand
feedback.WethankMalin Pinskyforhelpwith BRT modelling and
FredKrausforhissharingofdataonreleases. WealsothankAndy
RhyneforcuratingofLEMISdata.Finally,wethankmembersofthe
NationalSocio-EconomicSynthesisCenter(SESYNC)Pursuitonthe
pettradeandinvasivespeciesfortheirinsights.
AUTHORS’ CONTRIBUTIONS
O.C.S. and J.L.L. conceived the idea and wrote the manuscript.
O.C.S. c ollected and an alysed the data . All authors ap proved the
manuscript.
DATA ACCESSIBILITY
Data available via the Dryad Digital Repositor y https://doi.
org/10.5061/dryad.j2n732c(Stringham&Lockwood,2018).
ORCID
Oliver C. Stringham http://orcid.org/0000-0002-4224-7090
REFERENCES
Alle n,W.L .,Street ,S.E .,&C apellini,I.(2017).Fastlifehistorytraitspro-
moteinvasionsuccessinamphibiansandreptiles.Ecology Letters,20,
222–230.https://doi.org/10.1111/ele.12728
Blackburn,T.M.,Pyšek,P.,Bacher,S.,Carlton,J.T.,Duncan,R.P.,Jarošík,
V.,…Richardson,D.M.(2011).Aproposedunifiedframeworkforbi-
ologicalinvasions.Trends in Ecol ogy & Evolution,26,333–339.https://
doi.org/10.1016/j.tree.2011.03.023
Bomford , M., Krau s, F.,B arry, S. C. , & Lawrenc e, E. (200 9). Predict ing
establishment successfor alienreptiles and amphibians: A role for
climate matching. Biological Invasions, 11, 713–724. https://doi.
org/10.1007/s10530-0 08-9285-3
Capel lini, I., Bake r,J., A llen, W. L., Str eet, S. E., & Vend itti, C. (2 015).
Theroleoflifehistorytraitsinmammalianinvasionsuccess.Ecolog y
Letter s,18,1099–1107.htt ps://doi.org/10.1111/ele.12493
Capinha,C.,Seebens,H.,Cassey,P.,García-Díaz,P.,Lenzner,B.,Mang,T.,
…Essl,F.(2017).Diversit y,biogeographyandtheglobalflowsofalien
amphibiansandreptiles. Diversity and Distributions,23,1313–1322.
https://doi.org/10 .1111/dd i.12617
Cassey, P., Delean, S., Lock wood, J. L., Sadowski, J., & Blackburn, T.
M. (2018). Dissecting the null model for biological invasions: A
meta-analysis of thepropagule pressure effect. PLoS Biology, 16 (4),
e2005987.https://doi.org/10.1371/journal.pbio.2005987
Dorcas,M.E., Willson, J.D.,Reed, R.N.,Snow,R. W.,Rochford,M.R,
Miller, M. A ., … Hart , K. M. (2012). S evere mamma l declines co in-
cide with p roliferat ion of invasive Bur mese pyth ons in Everglad es
NationalPark.Proceedings of the National Academy of Sciences of the
United States of America, 10 9, 2418–2422. https://doi.org/10.1073/
pna s.111522610 9
Duggan,I.C., Rixon,C .A. M., & MacIsaac, H. J.(2006).Popularityand
propagule pressure: Determinants of introduction and establish-
ment of aquarium fish. Biological Invasions,8, 377–382.ht tps://doi.
org/10.1007/s10530-0 04-2310-2
EDDMapS.(2012).Early detection& distribution mappingsystem.The
UniversityofGeorgia.
Elith,J.,Leathwick,J.R.,&Hastie,T.(2008).Aworkingguidetoboosted
regressiontrees.Journal of Animal Ecology,77,802–813.https://doi.
org/10.1111/j.1365-2656.2008.01390.x
Essl, F., Dulli nger, S., Rabit sch, W., Hulme, P. E., Hülb er,K ., Jaroší k,
V., … Pyšek, P. (2011). Socioeconomic legacy yields an inva-
sion debt . Proceedings of the National Academy of Sciences of the
United States of America, 10 8, 2 03–207. http s://doi.or g/10.1073/
pnas.101172810 8
Fowler, A. J., Lodge, D. M., & Hsia, J. F. (2007). Failure of the
Lacey Act to protec t US ecosystems against animal invasions.
Frontiers in Ecology an d the Environment, 5, 353–359. https://doi.
org/10.1890/1540-9295(2007)5[353:FOTLAT]2.0.CO;2
Fujisaki,I., Hart,K .M.,Mazzotti,F.J.,Rice,K.G.,Snow,S.,& Rochford,
M. (2010). Riskassessment of potential invasiveness of exotic rep-
tiles imported to southFlorida.Biological Invasions,12,2585–2596.
https://doi.org/10.10 07/s10530 -0 09-9667-1
García-Díaz,P.,&Cassey,P.(2014). Patternsoftransportandintroduc-
tionofexoticamphibiansinAustralia.Diversity and Distributions,20,
455–466.https://doi.org/10.1111/ddi.12176
García-Díaz,P.,Ross,J.V.,Ayres,C.,&Cassey,P.(2015).Understanding
the biological invasion risk posed by the global wildlife trade:
Propagule pressure drives the introduction and establishment of
Nearctic turtles. Global Change Biology,21, 1078–1091. https://doi.
org /10.1111/gcb.12790
Hardin, S. (2007 ). Managing non-native wildlife in Flor ida: State per-
spect ive, polic y and pract ice. Managing Vertebrate Invasive Species,
14,43–52.
Holmberg, R. J., Tlust y, M. F., Futoma, E., Kaufman, L., Morris,
J. A., & Rhyne, A. L. (2015). The 800-pound grouper in the
room: Asymptotic body size and invasiveness of marine aquar-
ium fishes. Marine Policy, 53, 7–12. https://doi.org /10.1016/j.
marpol.2014.10.024
Hulme,P.E.(2015).Invasionpathwaysatacrossroad:Policyandresearch
challengesformanagingalienspeciesintroductions.Journal of Applie d
Ecology,52,1418–1424.https://doi.org/10.1111/1365-2664.12470
Hulme, P. E., Bac her,S ., Kenis, M., K lotz, S., Küh n, I., Minch in, D.,
… Vilà, M. (2008). Grasping at the routes of biological in-
vasions: A framework for integrating pathways into pol-
icy. Journal of Applied Ecology, 45, 403–414. https://doi.
org /10.1111/j.1365-266 4. 20 07.01442 .x
Hulme, P. E., Bacher, S., Kenis, M.,Kühn, I., Pergl, J., Pyšek, P.,… Vilà,
M.(2017).Blurringalienintroductionpathwaysriskslosingthefocus
|
9
Journal of Applied Ecology
STRIN GHAM ANd LO CKWOO d
oninvasivespeciespolicy.Conservation Letters,10,265–266.https://
doi .org/10.1111/conl .12262
Kraus, F. (2009). Alien reptiles and amphibians. Dordrecht, The
Netherlands:Springer.https://doi.org/10.1007/978-1-4020-8946-6
Krysko,K. L., Burgess, J. P.,Rochford,M. R.,Gillette, C. R ., Cueva, D.,
Enge, K. M ., … Nielsen, S. V. (2011). Verif ied non-indige nous am-
phibiansandreptilesinFlorida from1863through 2010: Outlining
the invasionprocess and identifying invasionpathways andstages.
Zootaxa,3028,1–64.
Leung,B.,&Mandrak,N.E.(2007).Theriskofestablishmentofaquatic
invasive species: Joining invasibility and propagule pressure.
Proceedi ngs of the Royal Society B, 274, 260 3–26 0 9. h t tps ://do i.
org/10.1098/rspb.2007.0841
Leung,B.,Roura-Pascual,N.,Bacher,S.,Heikkilä,J.,Brotons,L.,Burgman,
M.A.,…Sol,D.(2012).TE ASIngapartalienspeciesriskassessments:
A framework for best prac tices. Ecology Letters, 15, 1475–1493.
https://doi.org/10 .1111/el e.120 03
Lockwood, J.L.(2017). Exoticbirdsprovide uniqueinsight into species
invasions. Proceedings of the National Academy of Sciences of the
United States of America, 114, 9237–9239.ht tps://doi.org/10.1073/
pn as .17 12744114
Lockwood, J. L., Cassey, P., & Blackburn, T. (2005). The role of
propagule pressure in explaining species invasions. Tren ds in
Ecology & Evolution, 20, 223–228. https://doi.org/10.1016/j.
tree.2005.02.0 04
Mahoney,P.J.,Beard,K.H.,Durso,A.M.,Tallian,A.G.,LexineLong,A.,
Kindermann,R.J.,…Mohn,H.E.(2015).Introductioneffort,climate
matching a nd species trai ts as predictor s of global esta blishment
successinnon-nativereptiles.Diversity and Distributions,21,64–74.
https://doi.org/10 .1111/dd i.1224 0
Moorhouse, T. P., Balaskas, M., D’Cruze, N. C., & Macdonald, D. W.
(2017). Information could reduce consumer demand for exotic
pets. Conservation Letters, 10, 337–3 45. https ://doi.or g/10 .1111/
conl .12270
Reaser,J.K.,&Meyers,N.M.(20 07).Habitat titude:Gettingabackbone
aboutthepetreleasepathway.Managing Vertebrate Invasive Species,
40,63–71
Romagosa,C.M.(2014).Patternsoflivevertebrateimportationintothe
UnitedStates:Analysisofaninvasionpathway.InR.P.Keller,M.W.
Cadot te & G. Sandi ford (Eds). Invasive species in a globalized world:
Ecological, social , and legal p erspec tives on policy.Chicago,IL:Chicago
UniversityPress.
Schmidt,J.P.,& Drake,J. M.(2011).Whyaresome plantgenera more
invasivethanothers?PLoS ONE,6,e18654.ht tps://doi.org/10.1371/
journal.pone.0018654
Seekamp, E., Mayer,J. E., Charlebois,P.,& Hitzroth, G. (2016). Effects
of outrea ch on the preventi on of aquatic invasi ve species spre ad
amongorganism-in-tradehobbyists.Environmental Management,58,
797–809.https://doi.org/10.1007/s00267-016-0748-5
Shine, R . (2010). The ecologic al impact of invasive cane toads (Bufo
marinus) in Austr alia. The Quarterly Review of Biology,85, 253–291.
ht tps://do i.org /10.108 6/6 55116
Smith, K . F.,B ehrens, M., S chloegel, L . M., Marano, N ., Burgiel, S. , &
Daszak,P.(20 09). Reducing therisks of the wildlife trade. Science,
324,594–595.https://doi.org/10.1126/science.1174460
Strecker,A.L.,Campbell,P.M.,&Olden,J.D.(2011).Theaquariumtrade
asaninvasionpathwayinthepacificnorthwest.Fisheries,36,74–85.
https://doi.org/10.1577/03632415.2011.10389070
Stringh am, O. C., & Lock wood, J. L . (2018). Data from: Pe t problems:
Biological andeconomic factors that influence the release ofalien
reptiles and amphibians by pet owners. Dryad Digital Rep ository,
https://doi.org/10.5061/dryad.j2n732c
Su, S., Cassey, P., & Blackburn, T. M. (2016). The wildlife pet trade
as a driver of introduction and establishment in alien birds in
Taiwan. Biological Invasions, 18, 215–229. https://doi.org /10.1007/
s10530-015-1003-3
Su,S.,Cassey,P.,Vall-llosera,M.,&Blackburn,T.M.(2015).Goingcheap:
Determinants of bird pricein theTaiwanesepetmarket. PLoS ONE,
10,1–17.
Tingley,R.,Romagosa,C.M.,Kraus,F.,Bickford,D.,Phillips,B.L.,&Shine,
R.(2010).Thefrogfilter:Amphibianintroductionbiasdrivenbytax-
onomy,bodysizeandbiogeography.Global Ecology an d Biogeography,
19,496–503.
Vall-llose ra, M., Woolnou gh, A. P., Anderso n, D., & Cassey, P. (2017).
Improvedsurveillanceforearlydetectionofap otentialinvasivespe-
cies: Thealien Rose-ringedparakeetPsittacula krameriinAustralia.
Biological Invasions, 19, 1273–1284. https://doi.org/10.1007/
s10530-016-1332-x
Van Wilgen, N. J., & Richardson, D. M. (2012). The roles of cli-
mate, phylogenetic relatedness, introduction effort, and re-
productive traits in the establishment of non-native reptiles
and amphibians. Conservation Biology, 26, 267–277. https://doi.
org /10.1111/j.1523 -1739.2011. 018 04 .x
Whitf ield, P. E., Gardn er,T., Vive s, S. P., Hare, J. (20 02). Biologi cal in-
vasion ofthe Indo-Pacificlionfish Pterois volitansalongtheAtlantic
coastofNorthAmerica.Marine Ecology Progress Series,235,289–297.
https://doi.org/10.3354/meps235289
Wiles, G .J., Bart, J., Beck,R. E., & Aguon, C.F.(2003). Impacts of the
brown tre e snake: Patterns of d ecline and speci es persistence in
Guam’s avif auna. Conservation Biology, 17, 1350–1360. htt ps://doi.
org/10.1046/j.1523-1739.2003.01526.x
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Suppor tingInformationsectionattheendofthearticle.
How to cite this article:StringhamOC,LockwoodJL.Pet
problems:Biologicalandeconomicfactorsthatinfluencethe
releaseofalienreptilesandamphibiansbypetowners.J Appl
Ecol. 2018;00:1–9. https://doi.org/10.1111/1365-
2664.13237
