Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Sodium ion channel alkaloid resistance does not vary with toxicity in aposematic Dendrobates poison frogs: An examination of correlated trait evolution

  • Michael L. Yuan ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliations Department of Environmental Science, Policy, and Management, College of Natural Resources, University of California, Berkeley, California, United States of America, Museum of Vertebrate Zoology, University of California, Berkeley, California, United States of America

  • Ian J. Wang

    Roles Conceptualization, Funding acquisition, Resources, Supervision, Writing – review & editing

    Affiliations Department of Environmental Science, Policy, and Management, College of Natural Resources, University of California, Berkeley, California, United States of America, Museum of Vertebrate Zoology, University of California, Berkeley, California, United States of America


Spatial heterogeneity in the strength or agents of selection can lead to geographic variation in ecologically important phenotypes. Many dendrobatid frogs sequester alkaloid toxins from their diets and often exhibit fixed mutations at NaV1.4, a voltage-gated sodium ion channel associated with alkaloid toxin resistance. Yet previous studies have noted an absence of resistance mutations in individuals from several species known to sequester alkaloid toxins, suggesting possible intraspecific variation for alkaloid resistance in these species. Toxicity and alkaloid profiles vary substantially between populations in several poison frog species (genus Dendrobates) and are correlated with variation in a suite of related traits such as aposematic coloration. If resistance mutations are costly, due to alterations of channel gating properties, we expect that low toxicity populations will have reduced frequencies and potentially even the loss of resistance alleles. Here, we examine whether intraspecific variation in toxicity in three dendrobatid frogs is associated with intraspecific variation in alleles conferring toxin resistance. Specifically, we examine two species that display marked variation in toxicity throughout their native ranges (Dendrobates pumilio and D. granuliferus) and one species with reduced toxicity in its introduced range (D. auratus). However, we find no evidence for population-level variation in alkaloid resistance at NaV1.4. In fact, contrary to previous studies, we found that alkaloid resistance alleles were not absent in any populations of these species. All three species exhibit fixed alkaloid resistance mutations throughout their ranges, suggesting that these mutations are maintained even when alkaloid sequestration is substantially reduced.


Many species experience substantial variation in natural selection across their ranges leading to adaptive variation. In these scenarios, the relative costs and benefits of phenotypes under selection will vary across populations depending on local conditions, leading to local adaptation to optimize fitness tradeoffs [1,2]. Classically, variation in climate regimes often leads to widespread phenotypic variation tuned to local climate conditions (e.g. [3,4], and variation in predation pressure can drive the evolution of otherwise costly defenses in prey species (e.g. armor, chemical defense, etc.) [57] and resistance to defenses in predator species [810]. For example, predation pressure has driven the widespread evolution of chemical defenses [1116]. However, resistance to chemical defenses is associated with strong fitness tradeoffs in both prey, which must sequester and maintain toxins, and predator species [10,15,1719]. In both plants and sponges, increased chemical defenses are associated with decreased healing, growth, and recruitment rates [2022], and in garter snakes, Thamnophis sirtalis, resistance to TTX-defense in prey species is associated with decreased motor performance [23,24]. Thus, widespread intraspecific variation in toxicity is observed in a variety of taxa, potentially resulting from variation in local ecological conditions (e.g. predation pressure or environmental toxin availability) [5,22,25,26]. Such variation in toxicity may lead directly to variation in the strength of selection maintaining toxin autoresistance across populations.

Adaptation to transmembrane protein-targeting toxins is highly costly and constrained due to the essential biological function of these protein families [17,2731]. Thus, variation in toxin exposure often leads to variation in toxin resistance in response to fitness tradeoffs. A particularly well studied class of toxins are those which target the pore of voltage-gated sodium ion channels (NaVs)[17,32,33]. Populations of softshell clams, Mya arenaria, repeatedly exposed to saxitoxins (STX) have evolved resistance through a single NaV amino acid replacement [34,35]. Similarly, populations of T. sirtalis that prey upon highly tetrodotoxin (TTX) defended Taricha newts have independently evolved resistance at NaV paralogs [9,24,30]. In populations exposed to limited or no TTX, resistance is potentially deleterious due to reduced motor performance associated with TTX resistance [23,24]. Thus, population-level variation in resistance alleles throughout the range of T. sirtalis positively covaries with prey toxicity levels [5,9,23].

Poison frogs in the family Dendrobatidae have repeatedly evolved the capacity to sequester alkaloid toxins from their diets into specialized skin glands for chemical defense [16,36]. Three major classes of alkaloids found in poison frogs target NaV channels: batrachotoxins (BTX), histrionicotoxins (HTX), and pumiliotoxins/allopumiliotoxins (PTX/aPTX)[16]. The utilization of dietary alkaloids as chemical defense requires the congruent evolution of resistance to toxins from prey and autoresistance to a species’ own toxins once sequestered [15,19,37]. Several lineages of poison frog have convergently evolved resistance to alkaloid toxins in at least one sodium ion channel, NaV1.4, in congruence with alkaloids as chemical defense [38]. Encoded by the SCN4A gene, NaV1.4 is one of six NaV paralogs in amphibians [39]. Whereas several NaV paralogs are only expressed in the central nervous system and thus protected from many alkaloid toxins that might otherwise interact with them by the blood-brain barrier, NaV1.4 is readily exposed to all sequestered toxins due to its expression in skeletal muscle tissues [30]. Previous work has identified two amino acid replacements, A446D and I433V –occurring in Domain I Subunit 6 (DI-S6), which forms part of the NaV1.4 pore region—associated with alkaloid resistance in Dendrobates [38]. Following Tarvin et al. [38], we refer to amino acids by their homologous position in the Rattus norvegicus NaV1.4 protein. Docking models have shown the Aspartic Acid 446 (446D) amino acid residue inhibits binding of alkaloid toxins (i.e. PTX and HTX), although evidence for Valine 433 (433V) is more equivocal [38] and resistance has not been experimentally tested for these mutations. The influence of these amino acid replacements on BTX resistance is minor [40], but BTX defense is not observed in Dendrobates and is limited to their sister genus, Phyllobates [16,38,41]. Additionally, amino acid changes in the S6 pore region of NaV proteins may impose a fitness cost by substantially altering gating parameters, which can have consequences for normal channel function [42,43].

Despite widespread resistance mutations in other dendrobatids and both species being known to sequester alkaloid toxins [44], a recent study [38] reported an absence of alkaloid resistance in Dendrobates (= Oophaga) pumilio and D. (= Adelphobates) galactonotus NaV1.4 channels based on sequencing of single individuals in each species. If strong resistance to alkaloid toxins is costly to maintain, we should expect a pattern of covariation between alkaloid resistance alleles and toxicity, resulting in some populations having reduced Nav1.4 alkaloid resistance. Hence, this finding may reflect sampling drawn from populations with low alkaloid resistance allele frequencies rather than the absence of alkaloid resistance for the species entirely. Denser sampling of these and closely related species may reveal variation in alkaloid resistance traits across Dendrobates. Here, we examine whether the frequency of resistance alleles is reduced in lower toxicity populations relative to higher toxicity populations in three species of Dendrobates: D. pumilio, D. granuliferus, and D. auratus, focusing on two ecologically important substitutions (A446D and I433V) in the SCN4A gene that confer resistance to alkaloid toxins in other dendrobatids [38].

Several members of the histrionicus group, including D. pumilio and D. granuliferus (= O. granulifera), display dramatic variation in alkaloid toxicity throughout their ranges [26,4547]. In both species, variation in both total toxicity and alkaloid profiles are strongly correlated with a suite of associated phenotypes, such as aposematic coloration [46,47], suggesting multi-trait selection driven by toxicity as a predation defense. Dendrobates auratus is a member of the tinctorius group and was intentionally introduced to Manoa, Hawaii from Isla de Taboga, Panama in 1932 [48]. The Manoa population displays substantially reduced alkaloid profiles relative to populations in the native range [25], allowing us to test whether selection against costly alkaloid resistance due to reduction of dietary alkaloids is associated with a loss of resistance alleles.

Materials and methods

Sample collection

We assembled a collection of whole livers from 35 D. pumilio individuals collected in 2006–2007 [49], 11 D. granuliferus collected in 2008 [46], and 15 D. auratus collected in 2007 and 2015. The D. pumilio samples represent six localities: Isla Bastimentos, Panama (N = 9); Isla Pastores, Panama (N = 4); Isla Colón, Panama (N = 7); Tortuguero, Costa Rica (N = 8); Puerto Viejo, Costa Rica (N = 5); and Guapiles, Costa Rica (N = 2). Toxicity in Dendrobates has been quantified through several methods, most notably mouse sensitivity assays in which alkaloid extracts are injected subcutaneously into lab mice (similar in principle to commonly used LD50 assays) [46,47] and gas chromatography with mass spectrometry (GC-MS) to characterize alkaloid profiles [26,50]. Subcutaneous mouse assays have been criticized because mice are not clearly analogous to the natural predators of these species and because predators receive toxins orally rather than subcutaneously in nature [51,52]. These assays also include the effects of alkaloids that do not interact with NaV1.4 channels. However, estimates from mouse assays generally correlate with the diversity and concentration of major alkaloids in these species [26,4547]. Additionally, although not representative for the route of introduction in predators, for Dendrobates themselves subcutaneous injection is likely to expose toxins to more relevant tissues. Hereafter, we refer to mouse assays as ‘total toxicity’ and concentration of NaV1.4 acting alkaloids (i.e. PTX/aPTX and HTX) as ‘alkaloid profiles’. We discuss toxicity in the context of both datasets where literature is available for our sampled populations.

Dendrobates pumilio from the islands of Bocas del Toro, Panama include a wide variety of color morphs and toxicity levels [45,47,50,53]. Northwestern Isla Bastimentos represents one of the most toxic, whereas Isla Pastores and Isla Colón represent two of the least toxic populations in the region for both total toxicity and alkaloid profiles [45,47]. In fact, frogs on Isla Bastimentos are four to five times as toxic as frogs from Isla Colón in mouse assays [47]. The D. granuliferus samples represent four localities: Fila Chonta, Costa Rica (N = 2); Rio Savegre, Costa Rica (N = 1); Rio Baru, Costa Rica (N = 5); and Drake Bay, Costa Rica (N = 3). These localities represent the full range of color morphologies and associated total toxicity levels throughout the range of D. granuliferus [46] (Table 1). The Fila Chonta population is roughly 10% more toxic than the Rio Savegre population and about 60% more toxic than the Rio Baru and Drake Bay populations. Finally, we included tissues from D. auratus from five native localities and one introduced locality: Bananito, Costa Rica (N = 1); Puerto Viejo, Costa Rica (N = 2); Dominical, Costa Rica (N = 2); Uvita, Costa Rica (N = 2); Quepos, Costa Rica (N = 1); and Manoa, Hawai’i, USA (N = 7; Fig 1). We include sampling covering the Pacific (Quepos, Dominical, and Uvita) and Caribbean coasts (Bananito and Puerto Viejo) of Central America, which have different toxin profiles, and the less toxic introduced Manoa populations, which have reduced PTX/aPTX and lack HTX entirely [25].

Fig 1.

(A) Map of sampling localities. Relative haplotype frequencies for each sampling locality are denoted by pie charts. Colors for each haplotype correspond to haplotype networks displayed in B-D. (B-D) Haplotype networks for SCN4A in each species. Nodes sizes are scaled to represent the relative number of each haplotype and each unique haplotype is represented by a color. Photographs of each study species correspond to haplotype networks B-D respectively.

Table 1. Total toxicity assays from Wang 2011 [46] for D. granuliferus sequenced in this study.

Nuclear sequencing

We extracted whole genomic DNA from ethanol-preserved liver samples using a DNeasy Blood and Tissue Kit (Qiagen) following the manufacturer’s protocol. We PCR-amplified the Domain I Subunit 6 (DI-S6) of the NaV1.4 sodium ion channel encoded by the SCN4A gene. We targeted this locus using primers +869F (5’-CTGCAGGYAAAACCTACATGG-3’;) and +1032R (5’-GATGTTTCTTTAGCTGTTCC-3’) [38]. For all samples, we conducted 25μL reactions using 2% dimethyl sulfoxide, 1X standard Taq reaction buffer (NEB), 1.5mM MgCl2, 0.2mM dNTPs, 0.2μM final concentrations for each primer, and 0.625U Taq polymerase (NEB). We ran PCRs at an initial temperature of 94°C for 2min, followed by 30 cycles of 94°C for 30s, 51°C for 30s, and 72°C for 30s, followed by a final extension at 72°C for 5min. We then purified amplicons using ExoSAP-IT (USB). Next, we performed cycle sequencing reactions using Big Dye v3.1 sequencing chemistry (Applied Biosystems) and sequenced Sephadex G-50 cleaned products on an ABI Automated 3730xl DNA Analyzer (Applied Biosystems). We examined electropherograms by eye and assembled sequences in Sequencher v4.8. Finally, we deposited all newly generated sequences in EMBL-ENA (accession numbers: LT984926-LT984986).

NaV1.4 sequence analyses

We downloaded all available dendrobatid sequence data for NaV1.4 DI-S6 from Genbank (S1 Table). Additionally, we downloaded sequences from four non-dendrobatid hyloid frogs to serve as outgroups: Espadarana callistomma, Lithodytes lineatus, Bufo nubulifer, and Gastrotheca litonedis. We aligned previously available sequences with those generated in this study by amino acid sequence using MUSCLE [54]. To examine the evolutionary history of the NaV1.4 DI-S6 protein, we reconstructed a gene tree in BEAST2 [55]. We determined the best fitting partitioning scheme and associated models of nucleotide substitution using PARTITION FINDER 2 [56]. Specifically, we applied a K80+I model [57] to all codon positions. In BEAST, we ran two MCMC chains for 10,000,000 generations, sampling every 1,000 generations, and discarding 10,000 samples as burn-in. We assessed convergence of likelihood traces visually and by effective sample sizes (ESS) in Tracer v1.6.0 [58].

To examine SCN4A haplotype diversity across populations, we resolved haplotypes for heterozygous individuals with the PHASE algorithm in DNASP v5 [59]. We constructed median-joining networks [60] from phased haplotypes using POPART [61].

Ancestral state reconstruction

We reconstructed the mtDNA phylogeny of Dendrobatidae using previously published 12S, 16S, ND1, and ND2 sequences. We used the same sequences as Tarvin et al. [38], with the addition of previously published mtDNA sequences for D. auratus (HQ290980) and D. granuliferus (DQ502035) not included in that study. In BEAST, we applied a GTR+I+Γ model to all genes and ran two MCMC chains for 10,000,000 generations, sampling every 1,000 generations and discarding 10,000 samples as burn-in. As with our SCN4A gene tree, we assessed convergence in Tracer v1.6.0 [58]. Using the mtDNA tree, we estimated ancestral states for alkaloid resistance associated amino acid replacements I433V and A446D [38]. We compared models of character evolution using the Akaike information criterion (AIC). We applied the Mk1 model (Markov k-state 1 parameter, in which a single rate is applied to all transitions), a Symmetrical Rate model (forward and reverse transitions have equal rates), and an All-Rates-Different model and estimated marginal likelihoods in the ape package in R v3.3.3 [62]. Because the All-Rates-Different and symmetrical rate models are identical for binary characters, we tested only two models for I433V: Mk1 and the All-Rates-Different/Symmetrical Rate model.


All three Dendrobates species exhibited little to no intraspecific sequence variation in SCN4A across their ranges (Fig 1). We recovered no non-synonymous variation in any of the three species. All sequences of D. granuliferus were identical. All Bocas del Toro island populations of D. pumilio (Isla Bastimentos—high total toxicity, Isla Pastores—low total toxicity, Isla Colón—low total toxicity) share SCN4A haplotypes with each other and with mainland Costa Rican populations. The introduced Manoa population of D. auratus shares haplotypes with mainland Costa Rican populations in its native range. All three species were fixed for the A446D amino acid replacement, and D. auratus was also fixed for the I433V amino acid replacement.

Six dendrobatid sequences downloaded from Genbank (presented by ref. [38]) were identical despite being sequenced from species in different genera: Aromobates saltuensis (KT989176.1), D. galactonotus (KT989187.1), D. pumilio (KT989188.1), Phyllobates aurotaenia (KT989190.1), Colostethus fugax (KT989192.1), and Silverstoneia flotator (KT989194.1). Given the highly unlikely nature of six otherwise distantly related species sharing entirely identical sequences, particularly synonymous sites, we excluded these samples from our analyses. Our mtDNA tree was well supported and congruent with previous trees built from those data (Fig 2) [38,44]. The SCN4A gene tree was relatively poorly supported, with the posterior probabilities of over half the nodes less than 0.70. However, we still recovered a relatively similar topology to our mtDNA tree and previously published mtDNA trees (Fig 2) [38,44]. Although the internal relationships within each clade differ, the relationships between the Dendrobates-Phyllobates, Epipedobates, Hyloxalus, and Allobates clades were congruent between the mtDNA and SCN4A gene trees. For SCN4A, Ameerega parvula is sister to Epipedobates, with which it shares the A446D replacement, rather than the non-resistant members of the Ameerega-Colostethus clade (Fig 2). Additionally, D. (= Excidobates) captivus did not cluster with the Dendrobates-Phyllobates clade but rather diverges at the base of the gene tree (Fig 2).

Fig 2.

(A) Gene tree for NaV1.4 (SCN4A) DI-S6 inferred by BEAST. Branch lengths are transformed for ease of viewing and do not represent actual values. (B) mtDNA (12S, 16S, ND1, ND2) tree for Dendrobatidae. Nodes with posterior probability values above 0.70, 0.80, and 0.95 are denoted by shaded circles. Origins of resistance associated amino acid replacements, A446D and I433V, are denoted by boxes of D and V, respectively. Subsequent reversals to the ancestral state are denoted by strikethroughs. Lineages with alkaloid toxin defense are colored orange on both trees.

The Mk1, single transition rate, model was the best performing ancestral state reconstruction model for both amino acids. We recovered a single origin for both the I433V and A446D amino acid replacements in the common ancestor of Dendrobates and Phyllobates (Fig 3). We also recovered a subsequent reversal of I433V in the histrionicus group and a secondary D446E replacement in D. captivus (Fig 3).

Fig 3. Ancestral state reconstructions for alkaloid resistance-conferring amino acid replacements.

The underlying phylogeny was generated in BEAST using previously published 12S, 16S, ND1, and ND2 mtDNA sequences. Pie charts on nodes represent marginal likelihoods for ancestral character states.


We found that variation in NaV1.4 alkaloid resistance at the sequence level is not associated with intraspecific variation in toxicity for three species of poison frogs that exhibit dramatic variation in toxicity across their ranges. There is no loss of resistance associated alleles in Hawaiian D. auratus despite having been introduced to Manoa 85 years ago [48], having reduced diversity and concentrations of HTX and PTX/aPTX [25], and having short generation times [63,64]. Although, the Manoa population was founded by the Isla de Taboga population off the Pacific coast of central Panama [48], several Manoa SCN4A sequences are identical to individuals collected from mainland Costa Rica (Fig 1). This suggests that little to no structure in SCN4A exists across large portions of the D. auratus range in Costa Rica and Panama. Similarly, all SCN4A sequences from D. granuliferus were identical, despite individuals from populations in the southern part of the range (Rio Savegre and Fila Chonta) being significantly less toxic than individuals from northern populations (Rio Baru and Drake Bay; Table 1) in both toxicity assays and alkaloid composition [46]. These southern populations completely lack several toxins found in the north, including two pumiliotoxins, one histrionicotoxin, and one indolizidine alkaloid [46]. Thus, neither overall alkaloid toxin profile nor total toxicity is associated with genetic variation at SCN4A in D. granuliferus. Although we recover some synonymous site variation in D. pumilio, this variation is not associated with toxicity or geographic structure (Fig 1), and our sequences revealed no non-synonymous site variation across all populations. Both the highly toxic Isla Bastimentos and the substantially less toxic Isla Pastores and Isla Colón populations share SNC4A haplotypes with each other and with mainland populations. Because we do not have individual level toxicity data for D. pumilio and D. auratus, the possibility exists that the individuals sampled here are not fully representative of the average toxicity in their respective populations because toxicity has been shown to vary between individuals, sexes, and temporally [25,26,50]. However, intrapopulation variation in toxicity and alkaloid profile is substantially less than interpopulation variation [25,26,47,50], suggesting that comparisons between populations should be robust to individual-level differences. Thus, in all three species, our data show that reduced toxicity does not lead to the loss of potentially costly alkaloid resistance mutations.

Our data do not support our initial hypothesis that the previously reported absence of the A446D resistance mutation in D. pumilio NaV1.4 [38] can be explained by intraspecific variation in toxicity. On the contrary, we find no evidence to support the conclusion of previous studies [38] that some dendrobatid frogs lack the A446D replacement even though they sequester toxic alkaloids. All three of our study species were fixed for the A446D amino acid replacement, including D. pumilio and closely related D. granuliferus. Although we did not sample D. pumilio in Nicaragua, our Tortuguero locality in northeastern Costa Rica is <35 km from the Nicaraguan border. Additionally, the previously reported D. pumilio NaV1.4 DI-S6 sequence (KT989188), as well as those for D. galactinotus (KT989187) and Phyllobates aurotaenia (KT989190), are among the six identical sequences discarded from our analyses. These discarded sequences show a phylogenetic placement outside of known toxin sequestering clades (S1 Fig), suggesting they were not derived from Dendrobates or Phyllobates. Our data supports the hypothesis that all members of Dendrobates share a single ancestral origin of the A446D mutation (Fig 3) rather than the multiple origins previously posited [38]. Therefore, in D. captivus, the Glutamic Acid 446 residue likely arose from a secondary mutation from Aspartic Acid (D446E) rather than from the previously suggested Alanine ancestral state [38]. Furthermore, given the concerns about the previously reported NaV1.4 DI-S6 sequence for P. aurotaenia, the single A446D origin may extend to a single origin for both Dendrobates and its sister group, Phyllobates (Fig 3). However, we did not resequence members of Phyllobates in this study.

Additionally, our results suggest that D. auratus is fixed for the I433V amino acid replacement. The I443V replacement has also been reported in the closely-related D. tinctorius as well as the more distantly related D. captivus and P. terribilis. Our data support a single origin of I433V followed by a reversal in the histrionicus group (Fig 3). However, we cannot definitively rule out multiple origins of the I433V replacement. Alkaloid profiles indicate that D. auratus is likely to be more toxic than D. pumilio and D. granuliferus [41]. It is possible that the additional replacement in SNC4A from I433V confers additional resistance that allows for greater toxicity in species such as D. auratus and P. terribilis, the most toxic known dendrobatid [41,65]. However, the toxicity of D. captivus has not been characterized, and support for I433V contributing to PTX and HTX resistance is equivocal [38]. A446D and I433V provide some additional resistance to BTX when experimentally co-expressed in mammalian cells with additional replacements unique to P. terribilis, although their influence is minor [40].

If alkaloid resistance is costly to maintain in dendrobatid frogs, why do low toxicity populations maintain resistance phenotypes? One possible explanation is that sodium-ion channel (NaV1.4) alkaloid resistance is not actually costly to maintain. Although I433V and A446D shift gating properties in NaV1.4 when expressed in mammalian cells, whether this shift has functional consequences in Dendrobates is unknown [40]. Still, these mutations occur in the otherwise highly functionally conserved pore region of Nav1.4, and toxin resistance is associated with fitness costs in other vertebrates [17,28,66]. For example, TTX resistance alleles are associated with decreased motor performance in T. sirtalis [24]. However, experimental evidence has demonstrated that even adaptive phenotypes that are usually associated with fitness trade-offs can sometimes arise through underlying mechanisms with no associated cost [67]. Alternatively, the benefits of alkaloid resistance may outweigh the fitness costs even at low toxin concentrations, thus necessitating its maintenance in low toxicity populations. However, alkaloid resistance is maintained even in the Isla Pastores and Isla Colón populations of D. pumilio, which appears to be no more toxic in toxicity assays than non-toxic Allobates talamancae controls in mouse assays [47]. Costly alkaloid resistance genotypes may be maintained to protect against alkaloid toxins in prey items even when toxic prey are minor dietary components or when toxins are not actively sequestered by Dendrobates populations. Studies of prey communities, their alkaloid toxin profiles, and prey selection by Dendrobates are necessary to assess this hypothesis. Finally, the maintenance of alkaloid resistance at NaV1.4 may be a phylogenetic artifact [68]. If the A446D replacement was fixed in the ancestor of modern dendrobatines (Dendrobates and Phyllobates), reversal would require a de novo mutation, a stochastic event that simply may not have occurred in our study species. This scenario may explain the maintenance of the A446D replacement even if it is associated with a fitness tradeoff. Future studies on the fitness consequences of these mutations in various environmental contexts are needed to differentiate between these scenarios.


Contrary to expectations, we found no evidence for geographic variation in alkaloid resistance in the NaV1.4 voltage-gated sodium channel in D. pumilio, D. granuliferus, or D. auratus, even though each species displays dramatic variation in toxicity. Rather, our data suggest that these mutations are maintained even in populations where skin alkaloid profiles are substantially reduced, indicating either a lack of substantial fitness tradeoffs or the lack of reversal mutations. Further work is needed to determine the relative fitness costs, if any, for maintaining high level resistance despite low level toxicity. Additionally, we found no evidence for the previously reported absence of alkaloid resistance alleles in Dendrobates frogs [38]. Instead, our data support a single origin of the A446D amino acid replacement for the Dendrobates-Phyllobates clade and a single origin followed by a reversal for the I433V replacement. This single origin of initial alkaloid resistance is coincident with the hypothesized origin of alkaloid sequestration in the common ancestor of Dendrobates and Phyllobates frogs.

Supporting information

S1 Fig. Gene tree for NaV1.4 (SCN4A) DI-S6 inferred by BEAST including discarded sequences from Genbank.

All discarded sequences (shown in red) are identical in sequence and do not fall within known toxic clades of Dendrobatids. Posteriors above 0.80 are denoted.


S1 Table. Genbank accession numbers for previously published NaV1.4 DI-S6 and mtDNA (12S, 16S, ND1, and ND2) sequences used in this study.



We thank A.N. Wright for providing us with Hawaiian D. auratus tissues and MINAE (Costa Rica) and ANAM (Panama) for granting scientific collecting permits. Tissues were collected in accordance with protocols approved by the University of California, Davis IACUC. N. Graham provided some useful comments but mostly a considerate ear. We also thank L. Smith and the Museum of Vertebrate Zoology’s Evolutionary Genetics Lab for access to sequencing facilities.


  1. 1. Williams GC. Adaptation and natural selection: a critique of some current evolutionary ehought. Princeton University Press; 1966.
  2. 2. Reznick DN, Ghalambor CK. The population ecology of contemporary adaptations: what empirical studies reveal about the conditions that promote adaptive evolution. Genetica. 2001;112–113: 183–198. pmid:11838765
  3. 3. Phillimore AB, Hadfield JD, Jones OR, Smithers RJ. Differences in spawning date between populations of common frog reveal local adaptation. PNAS. 2010;107: 8292–8297. pmid:20404185
  4. 4. Fournier-Level A, Korte A, Cooper MD, Nordborg M, Schmitt J, Wilczek AM. A Map of local adaptation in Arabidopsis thaliana. Science. 2011;334: 86–89. pmid:21980109
  5. 5. Hanifin CT, Yotsu-Yamashita M, Yasumoto T, Brodie ED, Brodie ED. Toxicity of Dangerous Prey: Variation of tetrodotoxin levels within and among populations of the newt Taricha granulosa. J Chem Ecol. 1999;25: 2161–2175.
  6. 6. Marchinko KB. Predation’s role in repeated phenotypic and genetic divergence of armor in threespine stickleback. Evolution. 2009;63: 127–138. pmid:18803682
  7. 7. Van Buskirk J. The costs of an inducible defense in anuran larvae. Ecology. 2000;81: 2813–2821.
  8. 8. Duffy JE, Hay ME. Herbivore resistance to seaweed chemical defense: the roles of mobility and predation risk. Ecology. 1994;75: 1304–1319.
  9. 9. Brodie ED, Ridenhour BJ, Brodie ED. The evolutionary response of predators to dangerous prey: hotspots and coldspots in the geographic mosaic of coevolution between garter snakes and newts. Evolution. 2002;56: 2067–2082. pmid:12449493
  10. 10. Després L, David J-P, Gallet C. The evolutionary ecology of insect resistance to plant chemicals. Trends in Ecology & Evolution. 2007;22: 298–307. pmid:17324485
  11. 11. Roitberg BD, Isman MB. Insect chemical ecology: an evolutionary approach. Springer Science & Business Media; 1992.
  12. 12. Dumbacher JP, Beehler BM, Spande TF, Garraffo HM, Daly JW. Homobatrachotoxin in the genus Pitohui: chemical defense in birds? Science. 1992;258: 799–801. pmid:1439786
  13. 13. Clark VC, Raxworthy CJ, Rakotomalala V, Sierwald P, Fisher BL. Convergent evolution of chemical defense in poison frogs and arthropod prey between Madagascar and the Neotropics. PNAS. 2005;102: 11617–11622. pmid:16087888
  14. 14. Machado G, Carrera PC, Pomini AM, Marsaioli AJ. Chemical defense in harvestmen (Arachnida, Opiliones): do benzoquinone secretions deter invertebrate and vertebrate predators? J Chem Ecol. 2005;31: 2519–2539. pmid:16273426
  15. 15. Agrawal AA, Petschenka G, Bingham RA, Weber MG, Rasmann S. Toxic cardenolides: chemical ecology and coevolution of specialized plant-herbivore interactions. New Phytol. 2012;194: 28–45. pmid:22292897
  16. 16. Saporito RA, Donnelly MA, Spande TF, Garraffo HM. A review of chemical ecology in poison frogs. Chemoecology. 2012;22: 159–168.
  17. 17. Jost MC, Hillis DM, Lu Y, Kyle JW, Fozzard HA, Zakon HH. Toxin-resistant sodium channels: parallel adaptive evolution across a complete gene family. Mol Biol Evol. 2008;25: 1016–1024. pmid:18258611
  18. 18. Dobler S, Petschenka G, Pankoke H. Coping with toxic plant compounds—the insect’s perspective on iridoid glycosides and cardenolides. Phytochemistry. 2011;72: 1593–1604. pmid:21620425
  19. 19. Hanifin CT, Gilly WF. Evolutionary history of a complex adaptation: tetrodotoxin resistance in salamanders. Evolution. 2015;69: 232–244. pmid:25346116
  20. 20. Pawlik JR, Henkel TP, McMurray SE, López-Legentil S, Loh T-L, Rohde S. Patterns of sponge recruitment and growth on a shipwreck corroborate chemical defense resource trade-off. Marine Ecology Progress Series. 2008;368: 137–143.
  21. 21. Walters KD, Pawlik JR. Is there a trade-off between wound-healing and chemical defenses among Caribbean reef sponges? Integr Comp Biol. 2005;45: 352–358. pmid:21676780
  22. 22. Fine PVA, Miller ZJ, Mesones I, Irazuzta S, Appel HM, Stevens MHH, et al. The growth–defense trade-off and habitat specialization by plants in Amazonian forests. Ecology. 2006;87: S150–S162. pmid:16922310
  23. 23. Brodie ED. Evolutionary response of predators to dangerous prey: reduction of toxicity of newts and resistance of garter snakes in island populations. Evolution. 1991;45: 221–224. pmid:28564068
  24. 24. Geffeney S, Brodie ED, Ruben PC, Brodie ED. Mechanisms of adaptation in a predator-prey arms race: TTX-resistant sodium channels. Science. 2002;297: 1336–1339. pmid:12193784
  25. 25. Daly JW, Secunda SI, Garraffo HM, Spande TF, Wisnieski A, Nishihira C, et al. Variability in alkaloid profiles in neotropical poison frogs (Dendrobatidae): Genetic versus environmental determinants. Toxicon. 1992;30: 887–898. pmid:1523680
  26. 26. Saporito RA, Donnelly MA, Jain P, Martin Garraffo H, Spande TF, Daly JW. Spatial and temporal patterns of alkaloid variation in the poison frog Oophaga pumilio in Costa Rica and Panama over 30 years. Toxicon. 2007;50: 757–778. pmid:17706737
  27. 27. Aardema ML, Zhen Y, Andolfatto P. The evolution of cardenolide-resistant forms of Na+,K+-ATPase in Danainae butterflies. Molecular Ecology. 2012;21: 340–349. pmid:22126595
  28. 28. Feldman CR, Brodie ED, Brodie ED, Pfrender ME. Constraint shapes convergence in tetrodotoxin-resistant sodium channels of snakes. PNAS. 2012;109: 4556–4561. pmid:22392995
  29. 29. Zakon HH. Adaptive evolution of voltage-gated sodium channels: the first 800 million years. PNAS. 2012;109: 10619–10625. pmid:22723361
  30. 30. McGlothlin JW, Chuckalovcak JP, Janes DE, Edwards SV, Feldman CR, Brodie ED, et al. Parallel evolution of tetrodotoxin resistance in three voltage-gated sodium channel genes in the garter snake Thamnophis sirtalis. Mol Biol Evol. 2014;31: 2836–2846. pmid:25135948
  31. 31. Ujvari B, Casewell NR, Sunagar K, Arbuckle K, Wüster W, Lo N, et al. Widespread convergence in toxin resistance by predictable molecular evolution. PNAS. 2015;112: 11911–11916. pmid:26372961
  32. 32. Terlau H, Heinemann SH, Stühmer W, Pusch M, Conti F, Imoto K, et al. Mapping the site of block by tetrodotoxin and saxitoxin of sodium channel II. FEBS Letters. 1991;293: 93–96. pmid:1660007
  33. 33. Du Y, Garden DP, Wang L, Zhorov BS, Dong K. Identification of new batrachotoxin-sensing residues in segment IIIS6 of the sodium channel. J Biol Chem. 2011;286: 13151–13160. pmid:21303907
  34. 34. Bricelj VM, Connell L, Konoki K, MacQuarrie SP, Scheuer T, Catterall WA, et al. Sodium channel mutation leading to saxitoxin resistance in clams increases risk of PSP. Nature. 2005;434: 763–767. pmid:15815630
  35. 35. Connell LB, MacQuarrie SP, Twarog BM, Iszard M, Bricelj VM. Population differences in nerve resistance to paralytic shellfish toxins in softshell clam, Mya arenaria, associated with sodium channel mutations. Mar Biol. 2007;150: 1227–1236.
  36. 36. Santos JC, Coloma LA, Cannatella DC. Multiple, recurring origins of aposematism and diet specialization in poison frogs. PNAS. 2003;100: 12792–12797. pmid:14555763
  37. 37. Savitzky AH, Mori A, Hutchinson DA, Saporito RA, Burghardt GM, Lillywhite HB, et al. Sequestered defensive toxins in tetrapod vertebrates: principles, patterns, and prospects for future studies. Chemoecology. 2012;22: 141–158. pmid:22904605
  38. 38. Tarvin RD, Santos JC, O’Connell LA, Zakon HH, Cannatella DC. Convergent substitutions in a sodium channel suggest multiple origins of toxin resistance in poison frogs. Mol Biol Evol. 2016;33: 1068–1081. pmid:26782998
  39. 39. Zakon HH, Jost MC, Lu Y. Expansion of voltage-dependent Na+ Channel gene family in early tetrapods coincided with the emergence of terrestriality and increased brain complexity. Mol Biol Evol. 2011;28: 1415–1424. pmid:21148285
  40. 40. Wang S-Y, Wang GK. Single rat muscle Na+ channel mutation confers batrachotoxin autoresistance found in poison-dart frog Phyllobates terribilis. PNAS. 2017; 201707873. pmid:28874544
  41. 41. Daly JW, Brown GB, Mensah-Dwumah M, Myers CW. Classification of skin alkaloids from neotropical poison-dart frogs (Dendrobatidae). Toxicon. 1978;16: 163–188. pmid:635931
  42. 42. Yarov-Yarovoy V, McPhee JC, Idsvoog D, Pate C, Scheuer T, Catterall WA. Role of amino acid residues in transmembrane segments IS6 and IIS6 of the Na+ channel α subunit in voltage-dependent gating and drug block. J Biol Chem. 2002;277: 35393–35401. pmid:12130650
  43. 43. Wang S-Y, Bonner K, Russell C, Wang GK. Tryptophan scanning of D1S6 and D4S6 C-termini in voltage-sated sodium channels. Biophysical Journal. 2003;85: 911–920. pmid:12885638
  44. 44. Santos JC, Cannatella DC. Phenotypic integration emerges from aposematism and scale in poison frogs. PNAS. 2011;108: 6175–6180. pmid:21444790
  45. 45. Daly JW, Myers CW. Toxicity of Panamanian poison frogs (Dendrobates): some biological and chemical aspects. Science. 1967;156: 970–973. pmid:6023266
  46. 46. Wang IJ. Inversely related aposematic traits: reduced conspicuousness evolves with increased toxicity in a polymorphic poison-dart frog. Evolution. 2011;65: 1637–1649. pmid:21644954
  47. 47. Maan ME, Cummings ME, Adams AEDC, Shaw ERG. Poison frog colors are honest signals of toxicity, particularly for bird predators. The American Naturalist. 2012;179: E1–E14. pmid:22173468
  48. 48. Oliver JA, Shaw CE. The amphibians and reptiles of the Hawaiian Islands. Zoologica. 1953;38: 65–95.
  49. 49. Wang IJ, Shaffer HB. Rapid color evolution in an aposematic species: a phylogenetic analysis of color variation in the strikingly polymorphic strawberry poison-dart frog. Evolution. 2008;62: 2742–2759. pmid:18764916
  50. 50. Saporito RA, Donnelly MA, Garraffo HM, Spande TF, Daly JW. Geographic and seasonal variation in alkaloid-based chemical defenses of Dendrobates pumilio from Bocas del Toro, Panama. J Chem Ecol. 2006;32: 795–814. pmid:16718571
  51. 51. Weldon PJ. Poison frogs, defensive alkaloids, and sleepless mice: critique of a toxicity bioassay. Chemoecology. 2017;27: 123–126.
  52. 52. Bolton SK, Dickerson K, Saporito RA. Variable alkaloid defenses in the Dendrobatid poison frog Oophaga pumilio are perceived as differences in palatability to arthropods. J Chem Ecol. 2017;43: 273–289. pmid:28289966
  53. 53. Wang IJ, Summers K. Genetic structure is correlated with phenotypic divergence rather than geographic isolation in the highly polymorphic strawberry poison-dart frog. Molecular Ecology. 2010;19: 447–458. pmid:20025652
  54. 54. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004;32: 1792–1797. pmid:15034147
  55. 55. Bouckaert R, Heled J, Kühnert D, Vaughan T, Wu C-H, Xie D, et al. BEAST 2: a software platform for Bayesian evolutionary analysis. PLOS Computational Biology. 2014;10: e1003537. pmid:24722319
  56. 56. Lanfear R, Frandsen PB, Wright AM, Senfeld T, Calcott B. PartitionFinder 2: new methods for selecting partitioned models of evolution for molecular and morphological phylogenetic analyses. Mol Biol Evol. 2017;34: 772–773. pmid:28013191
  57. 57. Kimura M. A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J Mol Evol. 1980;16: 111–120. pmid:7463489
  58. 58. Rambaut A, Suchard MA, Xie D, Drummond AJ. Tracer v1.6 [Internet]. 2014.
  59. 59. Librado P, Rozas J. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics. 2009;25: 1451–1452. pmid:19346325
  60. 60. Bandelt HJ, Forster P, Röhl A. Median-joining networks for inferring intraspecific phylogenies. Mol Biol Evol. 1999;16: 37–48. pmid:10331250
  61. 61. Leigh JW, Bryant D. popart: full-feature software for haplotype network construction. Methods Ecol Evol. 2015;6: 1110–1116.
  62. 62. Paradis E, Blomberg S, Bolker B, Claude J, Cuong HS, Desper R, et al. ape: analyses of phylogenetics and evolution [Internet]. 2017.
  63. 63. Eaton TH. Notes on the life history of Dendrobates auratus. Copeia. 1941;1941: 93–95.
  64. 64. Summers K. Paternal care and the cost of polygyny in the green dart-poison frog. Behav Ecol Sociobiol. 1990;27: 307–313.
  65. 65. Daly JW. The chemistry of poisons in amphibian skin. PNAS. 1995;92: 9–13. pmid:7816854
  66. 66. Hilber K, Sandtner W, Kudlacek O, Schreiner B, Glaaser I, Schütz W, et al. Interaction between fast and ultra-slow inactivation in the voltage-gated sodium channel does the inactivation gate stabilize the channel structure? J Biol Chem. 2002;277: 37105–37115. pmid:12138168
  67. 67. Bennett AF, Lenski RE, Mittler JE. Evolutionary adaptation to temperature. I. Fitness responses of Escherichia coli to changes in its thermal environment. Evolution. 1992;46: 16–30. pmid:28564952
  68. 68. Stearns SC. The evolutionary links between fixed and variable traits. Acta Palaeontologica Polonica. 1993;38: 215–232.