Individual Variation in Alkaloid Content of Poison Frogs of Madagascar

15 Pages·2006·0.28 MB·English

JChemEcol(2006)32:2219–2233
DOI10.1007/s10886-006-9144-6
Individual Variation in Alkaloid Content of Poison Frogs
of Madagascar (Mantella; Mantellidae)
Valerie C. Clark&Valérie Rakotomalala&
Olga Ramilijaona&Leif Abrell&Brian L. Fisher
Received:6April2006/Revised:22May2006/
Accepted:9June2006/Publishedonline:30September2006
#SpringerScience+BusinessMedia,Inc.2006
Abstract Brightly colored Malagasy poison frogs, Mantella spp., sequester lipophilic,
basic alkaloids from arthropod prey for their own chemical defense. Consequently,
microsympatricpreydiversityisexpectedtoinfluencealkaloiddiversityobservedinpoison
frogs.Twenty-twospecimensofthreeMantellaspeciesfromfourlocalitiesinmoistforests
of southeastern Madagascar were analyzed individually via gas chromatography-mass
spectrometry, revealing that they contain over 80 known alkaloids. Frogs within a locality
possessed significantly similar alkaloid content and diversity, while frogs from areas that
varied in disturbance, elevation, and/or species showed greater differences. Based on
dietary data, the larger frog species Mantella baroni consumed more and larger prey, and
showed greater diversity in skin alkaloids than significantly smaller Mantella bernhardi.
Additionally, frogs from the most pristine locality had the greatest number of alkaloids,
whereasindividualsfromthemostdisturbedlocalitieshadtheleast.Inacomparisonoffrog
alkaloid profiles over a 10- to 14-yr period, alkaloid turnover, and thus presumably
alkaloid-sourcearthropodturnover,washighinadisturbedlocalityandlowinthepristine
primary forest locality. We demonstrate that the nonlethal transcutaneous amphibian
stimulator(TAS)iseffectiveforharvestingalkaloidsfrompoisonfrogs;futurestudiesusing
this device could obtain larger sample sizes without harming local frog populations.
ElectronicSupplementaryMaterial Supplementarymaterialisavailableintheonlineversionofthisarticle
athttp://dx.doi.org/10.1007/s10886-006-9144-6andisaccessibleforauthorizedusers.
V.C.Clark(*)
DepartmentofChemistryandChemicalBiology,CornellUniversity,Ithaca,NY14853,USA
e-mail:[email protected]
:
V.Rakotomalala O.Ramilijaona
DepartmentofAnimalBiology,UniversityofAntananarivo,Antananarivo101,Madagascar
L.Abrell
DepartmentofChemistry,UniversityofArizona,Tuczon,AZ85721,USA
B.L.Fisher
DepartmentofEntomology,CaliforniaAcademyofSciences,SanFrancisco,CA94103,USA
2220 JChemEcol(2006)32:2219–2233
Keywords Chemicaldefense.Allomones.Mantella.Chemicalecology.
Temporalvariation.Geographicvariation.Habitatdisturbance.
Antalkaloids.Alkaloidprofiles.Poisonfrogs
Abbreviations
TAS transcutaneous amphibian stimulator
Ampa Ampasimpotsy
Vato Vatoharanana
Saha Sahavondrona
Vohi Vohiparara
AMNH American Museum of Natural History
UADAB University of Antananarivo Department of Animal Biology
RA relative abundance in EI mode
SVL snout–vent length
Introduction
Frogs that are able to sequester alkaloids into skin from arthropod prey possess an
important evolutionary advantage; they not only afford themselves chemical protection
without expending energy to produce such defensive compounds, but also exploit a
nutrition niche that other species may reject as bitterly distasteful and dangerous (dietary
data not shown). Such ingested alkaloids are known nematicides, insecticides (including
antimosquito), and neurotoxins, and provide these frogs with protection against a wide
range of possible foes (Daly et al., 1999, 2005 and references therein). More than 800
lipophilic alkaloids from 24 classes, designated in boldface by molecular weight and
distinguishingletter(e.g.,251O; Fig. 1) (Daly et al., 2005), have been documented in certain
frog genera from five families. Most genera within these families do not have alkaloids, and
thusthetraitsforalkaloidsequestrationandstorageinskinglands,alongwithspecializationon
ants,haslikelyarisenmultipletimesinevolution(Clarketal.,2005andreferencestherein).
Such poison frogs include Mantella of Madagascar (Mantellidae), Dendrobates,
Epipedobates, and Phyllobates of Central and South America (Dendrobatidae), Melano-
phryniscusofSouthAmerica(Bufonidae),Pseudophyrneand,tosomeextent,Geocriniaof
Australia(Myobatrachidae;seeSmithetal.,2002),andtoalesserextent,oneLimnonectes
speciesofThailand(Ranidae;seeDalyetal.,2004,2005).Multiplefeedingexperimentswith
certain captive-born poison frogs revealed that: (1) dietary alkaloids are usually sequestered
unchanged;(2)somechemicalmodificationsofingestedalkaloidsexist;and(3)sequestered
alkaloidsare retainedin frogs’skin glands for several years (Smith et al.,2002; Daly et al.,
1997, 2003). Many of these “poison frog alkaloids” are also known from tropical and/or
temperateants,mites(Takadaetal.,2005),millipedes,beetles,andplants(reviewedinEisner
etal.,1978;KingandMeinwald,1996;Leclercqetal.,2000;Dalyetal.,2005;Laurentetal.,
2005).InMadagascar,asiphonotidmillipedeandthreeantsubfamilies,someofwhichwere
confirmed frog prey, contained alkaloids in common with sympatric Mantella(Clarketal.,
2005); however, putative sources of some of these same alkaloids occur in different ant
subfamiliesintheNeotropics(seeJonesetal.,1999).Furtherrelatedresearchonoccurrence
andpharmacologyofalkaloidsinpoisonfrogsandtheirpreyisreviewedinDalyetal.(1999,
2005),Saporitoet al. (2004),Clarket al.(2005) andDaly (2005).
JChemEcol(2006)32:2219–2233 2221
Fig. 1 Representative structures of select minor and major alkaloids detected in Mantella frogs of the
Ranomafanaregion,suspectedorconfirmedtobeofarthropodorigin
Daly (1982; see Table 22)and his colleagues (Myers and Daly, 1980; Daly et al., 1990,
1992,1994a,b,1996,2002)havedemonstratedthatthealkaloidcontentofpooledfrogskins
varybypopulation/locality.Myersetal.(1995)reportedvariabilitybetweenfrogs,withtwo
individualDendrobatesgranuliferusskinsateachofthreelocalesinCostaRica;also,atone
of these locales two individual and one 30-skin samples of microsympatric D. pumilio
tentativelyrevealedthatthetwospeciesaresimilarinalkaloidcontent.Saporitoetal.(2006)
alsodiscussvariationinalkaloidsrecoveredfrom70individualD.pumilioskins.Mebsetal.
(2005)mademethanolicextractsfrom81individualMelanophyrniscusmontevidensistoads
from Uruguay, to reveal by quantitative gas chromatography-mass spectrometry (GC-MS)
that there was variation among individuals and populations in the amount of both
pumiliotoxin 251D and thenonalkaloid phenolic hydroquinone.
We previously reported that within a locality, sympatric individuals of two Mantella
specieshadsimilaralkaloidprofiles(seesupportinginformationonlineandTable1inClark
et al. 2005). Here, we explore patterns of variation for >80 alkaloids detected in
individual Mantellapoisonfrogs.Individualvariationwithinoramongpoisonfrogspecies
2222 JChemEcol(2006)32:2219–2233
of a region may reveal important aspects of community ecology; thus, we report the
detailedalkaloidcontent(profiles)of22individualMantellafrogsofthreespeciessampled
from four localities in the Ranomafana region of southeastern Madagascar (Fig. 2),
alongside data on frog size and “snapshot” prey consumption. Alkaloid profiles from M.
baronicollectedattwooftheselocalitiesin1989and1993arecomparedtofrogsfromthe
2003 expedition.
Fig. 2 Mantella individuals were sampled in and near Ranomafana National Park at riparian locales
indicatedbysolidcircles(aftermapinBradt,1999)
JChemEcol(2006)32:2219–2233 2223
Methods and Materials
Field Collections
Materials and methods are described in detail in Clark et al. (2005). Briefly, riparian
Mantellamadagascariensis,M.baroni,andM.bernhardiwerecollectedinMarchandApril
2003 at localities in moist forests of the Ranomafana region in southeastern Madagascar
(Fig. 2). Localities included Ampasimpotsy (Ampa, disturbed forest fragments, 550 m; 21°
28.796′S,47°33.424′E),Vatoharanana(Vato,old-growthprimaryforest,1100m;21°17.444′
S, 47°25.569′E), Sahavondrona (Saha, disturbed roadside forest, 1100 m; 21°15.450′S, 47°
21.609′E), and both sides of the Kidonavo stream at Vohiparara (Vohi 1 and 2, disturbed
roadsideforest,1100m;21°13.587′S,47°22.193′E).Ampais25–34kmawayfromtheother
three localities, which are clustered 4–9 km from one another within Ranomafana National
Park (Fig. 2). In addition, temporal variations in the alkaloid content of M. baroni were
assessed with samples collected at Vato (December 1989, 10 combined skins) and Saha
(January 1993, 17 combined skins); these two samples were provided by John W. Daly
(National Institutesof Health), andpreviously described inDaly etal. (1996).
Sample Preparation and Analyses by Gas Chromatography-Mass Spectrometry
Eighteen Mantella frogs were skinned into methanol, and alkaloid fractions were made
followingthemethodsofDalyetal.(1994b).Theseandotherfrogvoucherspecimenswere
deposited and the stomach contents removed at the American Museum of Natural History
(AMNH, A168355–A168393) and the University of Antananarivo, Department of Animal
Biology(UADAB#1–8).Additionalexudateswerecollectedfromfourlivefrogsbyusing
a TAS (purchased from Jacqualine Grant, and described in Grant and Land, 2002) and a
methanol-laced Kim-wipe; these frogs were rinsed in water and released unharmed. Skin
alkaloid fractions and TAS extracts were individually analyzed for alkaloids on a
Micromass™ GC-time of flight (TOF)-MS in CI (with 100% NH ) and EI modes with a
3
ramp of 10°C/min from 100 to 280°C, with oven maintained at 100°C for the first 5 min
and at 280°C for the final 5 min. The MS libraries of Daly et al. (1999, 2005) were used
to identify over 80 known alkaloids in Mantella sampled at Ranomafana in 2003, and
the two M. baroni standards of Daly et al. (1996) provided retention times of known
alkaloidsforcomparison.Basedonlyupontheirrelativeabundance(RA)inEImode,these
alkaloidsarereportedinTable1astrace(RA<10%),minor(10%<RA<70%),ormajor
alkaloids (RA>70%; see chromatograms in Supplementary Information); although not
absolutely quantitative, any possible ionization biases should be consistent among all
samples tested.
Statistical Analyses
StatisticswereperformedwithSPSSv.13.At-testwasusedtocomparethemeanalkaloid
diversityofVohiM.baronisampledbyTASvs.full-skinextraction,andtocomparemeans
of both frog size [snout–vent length (SVL)] and number of prey between M. bernhardi
and M. baroni. A linear regression was used to evaluate the relationship between frog
size and number of prey within the Vohi locality. A partial correlation for all Mantella,
2224 JChemEcol(2006)32:2219–2233
b
bFamily bWorld Dalyetal. M,DM,DM,DM,D,BM,D,BDDM,DM,DM,D,BM,DM,DM,DM,DMM,DM,D?M,D?M,DMM×M,DM,D,BM,DM,D,B
7
U −−−−−−−−−−−−−−⊕⊕⊕−⊕−⊕−−−−−
8
U −−−−−−−−−−−−−−⊕ +⊗−⊕−⊕−−⊕−−
93 −−−−−−−−−−−−−−⊕ +−−⊕−⊕−−⊕−−
y
hardi mpots 92 −−−−−−−−−−−−−−−−−−⊕−⊕−−⊕−−
bern pasi 91 −−− +−−−−−−−−−−⊕− +−−−−−− +−−
m
M. A 90 −−−−−−−−−−−−−− +++−−−−−− +−−
89 −−−−−− +++++−− +⊗⊗−−−−−−−− +−
ana 88 −−−−−−− ++++−− +⊗⊗−−−−−−−−−−
n
a
har 87 −−−−−−− ++++−− +⊗⊗−−−−−−−− +−
o
ar Vat aD +−−−− +−−−− +−−−⊗⊗−−−−−−−−−−
c
gas U6 −−−−−−−−−−−−−− +⊗−−−− +−−−−−
a
d
Ma U5 −−−−−−−−−−−−−− +⊕−−−− +−−−−−
n a
n,southeaster M.baroni Sahavondron aD8457 −−+−−−−−+−−−−−−−−−−−+−−−−−−−−−−−−−−−−−−−−−−++⊕⊕⊕−−−−−+−−−−−−−++−−+−−−−−−−−−−−+
o
regi md.
ana M. U4 +−−−−−−⊕⊕ +− +−−⊕⊕−−−−−−⊕−−−
af
om 83 +−−−−−− +++− ++−⊕⊕−−−−−−−−− +
n
a
oftheR parara2 L4U3 −−−−−−−−−−−−−−−−−−−−−−−−−−−−⊕+++−−−−−+−+−−−−−−−−−−−−
auals Vohi L3 −−−−−−−−−−−−−− +⊕−−−−−−−−−−
d
vi 2
di L −−−−−−−−−−− +−− ++−−−−−−−−−−
n
i
og 1 82 −−−− +−−−−−− +−−⊕⊕−−−−−−−−−−
Mantellafr M.baroni Vohiparara U1U2 −−−−−−−− ++−−−−−−−−−−−− ++−−−−−⊕⊕+−−−−−−−−−−−−−−−−−− ++
n
etectedi Wcode 1B523C5B7B7E3A5A7A7J7N9B9I7A7B′7B″7B9F9L1I′1I23B3C3H
d M 55699990000000111111222222
s 11111112222222222222222222
d
oi
Table1Alkal Montellaspp. Locality Alkaloidclass PolyzonamineUnclassNicotinePrecoccinelline3,5-IPyrPip5,8-I5,8-I5,8-ITricyclicUnclass5,8-I5,8-I1,4-Q5,8-I5,8-I5,8-I5,8-I5,8-I5,8-I5,8-ISpiroP3,5-PIzidine3,5-P
JChemEcol(2006)32:2219–2233 2225
bal. B? B B BB? B
bWorld Dalyet M,D,××M,D,M,DM,D?M,DM,DM,DM,D,×M,DM,D,M,D,DMMM,DM,DMMMMM,DM,DM,DM,DM,D,M,DM,DM,DMM,D
7
U − ++−−−−⊗−−−−−−−−−−−−−⊕⊕−⊕−−−−−−⊕−
8
U − +−−−−−⊕−−−−−−−−−−−−− ++− +−−−−−−⊕−
93 − ++−−−−⊕−−− +−−−−−−−−−⊕⊕−−−−−−−−⊗−
y
s
pot 92 −−−−−−−⊗−−−−−−−−−−−−−−−−−−−−−−−⊗−
m
asi 91 +⊕−−−−−⊕−−−−−−−−−−−−−−−−−−−−−−−⊗−
p
m
A 90 ++−−−−−⊗−−−−−−−−−−−−−−⊕−−−−−−−−⊗−
89 −−−−⊕−−⊗−−−⊗−−−−−−−−⊕−⊕−− ++⊗−−−⊗−
ana 88 −−−−⊕−−⊕−−−⊗−−−−−−−−⊕−⊕−− +−⊕−−−⊕−
n
a
har 87 −−−−⊕−−⊕−−−⊗−−−−−−−−⊕−⊕−− +−⊕−−−⊕−
o
Vat aD −−−−⊕−−⊕−−−⊗−−−−−−−−⊕−⊕−−−− +−−− +−
6
U −−−−⊕−−−− +−−−−−−−−−−−−−−−−−−−−−⊗−
5
U −−−−⊕−−−− +−−−−−−−⊕−−−−−−−−−−−−−⊗−
a
dron 57 −−−−⊗ +− +−−−−−−−−− ++−−−−−−−−−−−−⊕−
von 84 −−−−⊗ +−−−−−−−−−−−⊕− +− +− +++−−−−−⊕−
a
h
Sa aD −−−−⊕−−⊕− +− +−−−−− +−−−−−−−−−⊕−− +⊕ +
4
U −⊕ +−−−−−−⊕⊕−⊕⊕ +⊕⊕−−−−−−−−−−⊕−−− +−
83 −⊕ +−⊕−−−− ++−−−−−−−−−−−−− +−−⊕− +−⊕−
2 U3 +++− +− +−−−−−−−−−−−−−−−−−−−−⊕−− ++−
a
parar L4 −−−− +−−− +−− +−−−−−−−−−−−−−−−⊕ +− ++−
hi
Vo L3 −−−−⊕−−⊕−−−−−−−−−−−−−−−−− +−⊕−−− +−
L2 −−−− +−− +−−− +−−−−−−−−−−−−− ++⊗−−−⊕−
1 82 −−−−⊕−−⊕−−− +−−−−−−−−−−−−− +−⊗ +−−⊗−
a
arar U2 −−− +⊕ +−−−−− +−−−−−⊕−−−−−−−−−⊗−−−⊗ +
p
hi
Vo U1 −⊕ +− +−−−−−− +−−−−−−−−−−−−− +−⊗−−−⊕ +
e
d
Wco ′3H″3H‴3H3I1A′1A1J3A5R6′67A7G′7G89K′9K1F3B3C3D5B5C5E7E9A′9A1D1H1M1N1O1R
M 222233333333333334444444444555555
222222222222222222222222222222222
s
s
a
cl X
Locality Alkaloid 3,5-P3,5-P3,5-P5,8-I1,4-Q1,4-QUnclass1,4-QUnclassSpiroPSpiroPPTX3,5-P3,5-PSpiroP3,5-P3,5-P5,8-I5,8-I5,8-I5,8-I5,8-I5,8-IDHQ5,8-I3,5-I3,5-IPTXDeoxyPT5,6,8-I5,8-I3,5-PhPTX
2226 JChemEcol(2006)32:2219–2233
bFamily bWorld bDalyetal. M,DM,DM,DMMM,D,BM,D,B,MyoM,DMMMM,DM,DM,DDMM,D,BMM,DMMM,DDM,DM,DM,D
7
U −−−−−−−−−−−−−−−−−−−−−−−−−−
8
U −−−−−−−−−−−−−−−−−−−−−−−−−−
3
9 −−−−−−−−−−−−−−−−−−−−−−−−−−
y
hardi mpots 92 −−−−−−−−−−−−−−−−−−−−−−−−−−
bern pasi 91 −−−−−−−−−−−−−−−−−−−−−−−−−−
m
M. A 90 −−−−−−−−−−−−−−−−−−−−−−−−−−
89 −⊕⊕−−−−−−−−⊕⊗−−−−−−⊕− +−⊕⊕⊕
ana 88 − +⊕−−−− +−−− +⊕−− +−−− +−−− ++⊕
n
a
har 87 −− +−−−− +−−−⊕⊕−− +−−− +−−−−−⊗
o
Vat aD − ++−−−− +−−− +−−−⊕− +− +− +− ++⊕
6
U −−−−−⊕−−− +−−−−−−−−−−−−−−−⊕
5
U −− +−−−−−−−−− +−−−−−−−−−−−−⊕
a
aroni vondron 8457 −−−−−+−−−−⊕+−−−−−−⊕+−+−+⊕−−−−−−−−−−−−−−−−−−−−−−−−−⊕−
b a
h
M. Sa aD +−−−⊕⊗−−⊕⊕ ++−−⊕ +−−−⊕−⊗− +− +
d.
m
M. U4 −−− +−⊗⊗−−−−− +⊕−−−−−− +−⊕−− +
83 −− +−−⊗−−−⊕−− +⊕−−⊕−−−−−−−−⊗
2 U3 −−−−−⊗−−−⊕− +⊗−−−−−−−−−−−−⊕
a
parar L4 −− +−−⊕−−−⊕− +⊕−−−−−−−−−−−− +
hi
Vo L3 −−−−−⊕− +− +−⊕−−−−−−−−−−−−−−
L2 −−−−−⊕−−−⊕−⊕⊕−−−−−−−−−−−−−
1 82 −−−−−⊗− +−⊗− +⊕−− +−−− +−−−−−⊗
roni arara U2 −−−−−⊗−−−⊗− +⊕−−−−− +⊕−−−−−⊗
a p
M.b Vohi U1 −−−−−⊗−−−⊗− ++−− +−−−⊕−−−−−⊕
e
d
o
Wc 2A5B7D1C5F5N7C1B1F3A′3A5C′5C5E1G1E1G3B3C3D5B7A7D″7F‴7F7G
d) M 2525252626262627272727272727282929292929303030303030
e
u
n
(conti spp. class XX X X
Table1 Montella Locality Alkaloid SpiroPIzidine1,4-QTricyclicUnclasshPTXPTXUnclass3,5-I5,6,8-I5,6,8-I3,5-I3,5-I5,6,8-IDeoxyPTDeoxyPTPTXDeoxyPT5,6,8-IDeoxyPTPTXPTXAPTXPTXPTXPTX
1 2
3 3
1. 1.
t t
JChemEcol(2006)32:2219–2233 2227
bMontellaspp.M.baroniM.md.M.baroniM.bernhardiFamily bLocalityVohiparara1Vohiparara2SahavondronaVatoharananaAmpasimpotsyWorld aabAlkaloidclassMWcodeU1U282L2L3L4U383U4D8457U5U6D87888990919293U8U7Dalyetal. ⊗⊗⊗⊕⊗⊕⊗⊗⊕⊕−⊗⊕⊕−−−−−PTX309A+++++M,D−−−−−−−−−−−−−−⊕−−−−−−−−−PTX309CD−−−−−−−−−−−−−−⊕⊕⊕⊕⊕aPTX323B+++++M,D,B,Myo−−−−−−−−−−−−−−−−−−−−−hPTX323E+++M−−⊕−−⊕−⊕⊕⊕⊕⊗aPTX325A++++++++++++M,D,MyoTotal#ofalkaloids20232115131917262830251210112726293111118141414Avg.#alkaloids/frog/locality19.8±3.420.2±4.920.6±6.314.5±7.028.7±2.512.0±2.4Total#ofalkaloids/locality31(Vohi1)59(both)49(Vohi2)3028273320K-meansClusterGroupAAAAAAABBACCCCDDDDCCCCCCTAS(T)orSkin(S)SSSTTTSSSSSTSSSSSSSSSSSSc474833ndndnd216613nd27234130nd3040101627167364No.preyrecovered–c(mm)272424242725223022nd27252228nd262427172220212219Snoutventlength ’−⊕M.md:Mantellamadagascariensis(singlespecimen=U4);D:Dalyssamples;:notdetected;+:tracealkaloid[(<10%relativeabundance(RA)inEImode];:minor⊗′alkaloid(10%<RA<70%);:majoralkaloid(RA>70%);nd:nodata;primesymbol()indicatesisomerofadifferentGCretentiontime.Alkaloidclasses(seeFig.1):#,#substituted...P:pyrrolizidine;I:indolizidine;Q:quinolizidine;Pyr:pyrrolidine;Pip:piperidine.(a,h)PTX:(allo,homo)pumiliotoxin;SpiroP:spiropyrrolizidine;DHQ:decahydroquinoline;Unclass:noclassassigned.IndividualMantellafrogsarecodedwithvoucheridentificationnumbers,asU#forUADABdepositedMantella#s1-8,andas##forAMNHdepositedfrogsA1683##(A168357-A168393).L#referstolivefrogswhoseexudateshadbeencollectedwithatranscutaneousamphibianstimulator,photographed,andthenreleasedunharmed.aOuranalysesofM.baronisamplesprovidedbyDalyforSaha(January1993,17skins)andVato(December1989,10skins);seeDalyetal.(1996).—bDatafromDalyetal.(1996,1999,2005)thelatterincludesdatafromClarketal.(2005).Occurrenceofisomerswaspooled.Frogfamilies:M,Mantellidae(Madagascar);D,Dendrobatidae(CentralandSouthAmerica);B,Bufonidae(SouthAmerica);Myo,Myobatrachidae(Australia);×,notincludedinDalyetal.(2005).cAM.bernhardispecimen(A168358)fromAmpanotanalyzedforalkaloidsmeasured18mm,and10arthropodswererecoveredfromitsstomach.
2228 JChemEcol(2006)32:2219–2233
controlled for by locality, was used to evaluate the relationship of SVL with both the
number of prey and individual frog skin-alkaloid diversity. One-way analysis of variance
(ANOVA)wasusedtocomparemeanalkaloiddiversity perfrogamongallfourlocalities,
among the three localities of constant (1100 m) elevation, and between frogs of the two
elevation groups. K-Means cluster analysis was performed on 22 Mantella frogs by using
all alkaloids requesting four clusters, and was followed by a cross-tabulation between the
cluster membership and the localities.
Results and Discussion
Variation in Alkaloid Content of Individual Frogs, within and among Localities
ClusteranalysisrevealedthatthethreeM.baroniand1989standardfromVatorepresenta
distinct group based on the specific alkaloid content of the resident individual frogs (see
ClustergroupsatbottomofTable1),indicatingthatfrogswithinalocalityaremoresimilar
to one another compared to frogs from different localities. Seven of nine frogs from both
sides of the Kidonavo stream at Vohi clustered in another group of individuals, indicating
littlevariationatthisfinespatialscale(Table1).TheothertwoVohifrogsclusteredinathird
uniquegroup,demonstratinglittledifferenceinalkaloidprofilesofmicrosympatricMantella
species. Frog individuals from the two most disturbed locales clustered exclusively in
another group, including all four frogs from moderately disturbed Saha and all six frogs
fromthedisturbedforestfragmentsofAmpa;thesetwolocalitiesalsohadthelowestmean
number of alkaloids per frog and the six individuals with the fewest alkaloids (Table 1).
Outofthe91alkaloidsinTable1,26weredetectedinonlyasinglefrogindividual,16were
detectedinmultiplefrogsfromasinglelocality,and11weredetectedinmultipleindividuals
fromtwolocalities.Vohi1and2areseparatedonlybythe2-to4-m-wideKidonavostream,
and frogs from these sites contained many alkaloids in common (e.g., 207N, 251D, 265N,
273A);21alkaloidswerepresentonbothsidesoftheKidonavostream,andtheother38were
onlydetectedinfrogsfromeitherVohi1orVohi2(Table1; ofthese38alkaloids,25were
foundonlyinasinglefrog).Ofthe91alkaloids,26weredetectedinonlyoneofthevarious22
frogssampled,andmayrepresentrareprey.ThefactthatVohisamplescontain25ofthese26
unique(inoneof22frogs)alkaloidssuggeststhisintermediatelydisturbedsitemayhavean
elevatednumberofrarearthropod-sourcespecies.Alkaloidsappearinginmorethanonefrog
thatarelimitedtoonelocalityinclude:217Band223BatAmpa; 151B,195B, 207N,251R,
and 275E at Vohi; 273A′ at Saha; and 207J, 209I, 223C, 223M, 243D, 255B, 307F″, and
307F′″ at Vato. These results also suggest that frogs within a locality are more similar in
termsofalkaloidcontentcomparedtofrogsfromdifferentlocalities.Alkaloidsthatappearin
the majority of frogs from only two localities in 2003 include 203A, 205A, 207A, 219F,
221I, 223H″, 241F, 245C, 265N, 291E, and 293D (Table 1).
The distribution and amount of particular alkaloids in individual Mantella frogs should
reflect how abundant the leaf-litter source arthropod is in the environment. Although frog
sequestration efficiency varies among alkaloid classes (see Daly et al., 2003), such
sequestration biases should be constant among Mantella sampled in this study. For
example,alkaloidssuchas251Oand217Bwerefoundinallormostofthe22individuals
(Table1,Fig.1);thesealkaloidsarelikelycommoninfrogsbecausethesourcearthropodis
particularly abundant in the region, so it was not surprising that the endemic ant source
species for these common alkaloids was discovered (Clark et al., 2005). As-yet-
undetermined arthropod sources for 217A, 233A, and 325A are also likely widely distrib-